FAQ: the platform, its data, models and evidence

Question 1

What is Opulence Alpha?

Accepted Answer

Opulence Alpha is non-custodial investment research software for US-listed stocks. It forecasts returns with an interval around each forecast, estimates fair value, reads the market regime across nine axes, tests strategies walk-forward and then measures what actually happened. The Regime Radar, the Fair Value page, the weekly Golden Tickets record and 24 tested market beliefs are free to read. It runs machine-learning models, not generative AI, and it never holds money or places orders.

Question 2

Who is Opulence Alpha for?

Accepted Answer

It is built for investors who want evidence behind a decision: self-directed investors with a process (Investor plan); advisers, family offices and quantitative researchers who test their own policy against history (Professional); and desks that need a shared pool, programmatic access and an audit trail (Institutional). Reading research already produced is free on every plan, and the 30-day trial of every Professional feature needs no card. It is not a service that manages money for you.

Question 3

Is Opulence Alpha investment advice?

Accepted Answer

No. Opulence Alpha provides quantitative research and software tools; nothing on the site is investment advice, a recommendation, or an offer to buy or sell any security. Its forecasts and fair values are model estimates, and its studies are measured tendencies across many stocks, published with their limits. The platform does not decide for you: decision and execution authority is always yours. Past performance does not guarantee future results, and all investing involves risk of loss.

Question 4

Does Opulence Alpha hold my money or connect to my broker?

Accepted Answer

No. Opulence Alpha is non-custodial research software: it does not hold client money or assets, does not connect to your broker and does not place orders. The platform simulates the portfolio you pick at the risk profile you chose and reports the rebalances it simulated; test portfolios hold no money. Acting on any of it is your decision, placed through your own broker.

Question 5

Where is Opulence Alpha based?

Accepted Answer

Opulence Alpha is based in Istanbul, Turkey, at Maslak Mahallesi, Büyükdere Caddesi, Sarıyer. The console runs on Google Cloud Run in a United States region, and its databases run on Google Cloud as well. The public site and the console are available in English and Turkish, and messages sent through the contact form are read and answered by a person.

Question 6

What data does Opulence Alpha use?

Accepted Answer

Opulence Alpha uses daily prices for US-listed stocks, adjusted for splits and, where the source records them, dividends; company filings from SEC EDGAR, including Form 4 insider trades (from 2003) and Schedule 13D activist stakes; macroeconomic series from the St. Louis Fed's FRED database; FINRA short interest (from 2021) and daily short-sale volume (from 2009); and analyst rating and price-target actions (from 2011). FinBERT, a finance-specific classifier, scores the tone of SEC filings as one input feature.

Question 7

How many companies does Opulence Alpha cover, and since when?

Accepted Answer

The research universe is 1,767 US-listed common stocks across 11 sectors; 422 of them have since delisted and stay in the studies with their history. The studies observe it every Wednesday from Jan 4, 1995 to Aug 19, 2026: 1,634 Wednesdays spanning 7,959 trading sessions. They start in 1995 because adjusted closes before then are unreliable in the source data. On Sep 25, 2026, the daily Fair Value page covered the 1,266 US companies in coverage that were listed that day; 1,227 of them had a fair value.

Question 8

How does Opulence Alpha handle survivorship bias?

Accepted Answer

Companies that failed or were acquired stay in. Every company in the S&P 500 at any time since 1996 whose price history could be recovered is included with its full history: 1,027 companies, 413 since delisted. Two gaps remain: 137 former members are missing, mostly companies that left the index before 2010, and the 740 non-index companies were chosen from those listed at construction, so smaller companies that failed or were acquired earlier are under-represented. Lead results are therefore re-measured on each date's S&P 500 members.

Question 9

What does point-in-time mean at Opulence Alpha?

Accepted Answer

It means every test uses only what was known on that date. Fundamentals are read as of their SEC filing dates; insider and analyst readings start the session after the filing or action was public; economic releases enter on the day they were published; and short interest is dated to its estimated publication day, eight business days after the settlement date it describes. Signals are also recomputed over the past with today's definitions, so a study is a measured tendency, not the record of a strategy run live.

Question 10

Does Opulence Alpha cover markets outside the US?

Accepted Answer

No. Coverage is US-listed common stocks on the NYSE and Nasdaq, across the 11 GICS sectors; funds, ETFs, trusts, preferred shares, warrants and units are excluded. Shares listed only outside the US, for example on Borsa Istanbul or on European and Asian exchanges, are not covered, and the Regime Radar reads the US market. The Turkish-language pages present the same US research in Turkish; they do not cover Turkish companies.

Question 11

Does Opulence Alpha use ChatGPT, Gemini or Claude?

Accepted Answer

No. The platform calls no ChatGPT, Gemini or Claude service, and no generative language model writes, summarises or decides anything in its research. Its forecasts come from machine-learning models trained on market, filing and macroeconomic data: gradient-boosted decision trees, quantile models and LSTM neural networks. The only language model in the system is FinBERT, a classifier that scores the tone of SEC filings as one input among many.

Question 12

Is there an AI chatbot or assistant on Opulence Alpha?

Accepted Answer

No. Opulence Alpha has no chatbot or conversational assistant, and the platform generates no text with an AI model. Its research arrives as numbers that can be checked: return forecasts with a range around them in the signed-in console, daily regime readings, fair-value estimates, and the studies behind them. The methods are described on the Technology, Regime Radar and Studies pages.

Question 13

What AI models does Opulence Alpha use?

Accepted Answer

Opulence Alpha uses predictive machine-learning models: gradient-boosted decision trees (CatBoost), LSTM neural networks (PyTorch), and XGBoost quantile models whose forecast ranges are calibrated by conformal quantile regression (CQR). Bernstein Online Aggregation combines their forecasts. Market regimes come from a percentile classifier on each axis, fused by a hidden semi-Markov model. These are published methods built with open-source libraries, trained on the platform's own dataset, which is built to be point-in-time.

Question 14

Does Opulence Alpha use any language model at all?

Accepted Answer

One, and it writes nothing. FinBERT (ProsusAI/finbert) is an open-source classifier built for financial text. Opulence Alpha uses it to label the tone of SEC filing text as positive, negative or neutral, and those scores feed the sentiment inputs available to the forecasting models. It does not generate text, summarise documents or make decisions.

Question 15

How does Opulence Alpha learn from its results over time?

Accepted Answer

Every trading day, as earlier forecasts reach their horizon, each model is scored on how well it ranked stocks against how they actually moved. Bernstein Online Aggregation then shifts weight toward the models that have been ranking better, separately for each horizon. Older evidence fades gradually, and a small floor keeps every model's learned weight above zero, so a model that starts predicting well again can regain influence. The models themselves change only when they are retrained.

Question 16

How often are the models retrained, and which version is used?

Accepted Answer

The 1-day and 5-day models are scheduled to retrain every Saturday; on the first Saturday of each month, every model at every horizon (1, 5, 21 and 63 days) retrains, along with the interval models. The newest trained model becomes the champion automatically. A separate statistical gate on held-out data decides whether that champion carries trading weight: a model that fails it is still scored every day but does not shape the published forecasts.

Question 17

What is the difference between predictive AI and generative AI?

Accepted Answer

Generative AI, such as ChatGPT, produces new text or images from patterns in its training data. Predictive AI learns from past data to estimate a quantity: at Opulence Alpha, a stock's return over a 1-, 5-, 21- or 63-day horizon, with a range around it. The practical difference is accountability. A predictive forecast matures and can be scored against what actually happened, and that daily scoring is how the platform weighs its own models.

Question 18

How does Opulence Alpha show how uncertain a forecast is?

Accepted Answer

Forecasts are published as a range, not only a single number: a central estimate for each horizon with an interval around it, produced by conformal quantile regression. The interval is calibrated on past forecast errors and assumes no particular shape for returns; a forecast without a calibrated interval carries none. How often outcomes land inside the interval is monitored, alongside calibration and CRPS. A forecast is a model estimate with stated uncertainty, not a price target.

Question 19

How does Opulence Alpha detect market regimes?

Accepted Answer

Regime Radar reads nine axes: volatility, sentiment, liquidity, credit, recession risk, the fundamental cycle, inflation, real yields and monetary policy. Each axis's state comes from where today's readings sit in their own history, against percentile thresholds set for that axis. A hidden semi-Markov model fuses the nine into one composite that accounts for how long regimes tend to last. A hidden Markov model and a gradient-boosted classifier are trained per axis only as cross-checks.

Question 20

How is Opulence Alpha hosted?

Accepted Answer

Opulence Alpha is hosted on Google Cloud. The website and console run on Google Cloud Run in a United States region. The research warehouse and the transactional database also run on Google Cloud, reached over a private network and not exposed to the internet. The research services run as separate containers, scheduled by Apache Airflow. The privacy page states exactly what is, and is not, claimed about encryption.

Question 21

Does Opulence Alpha send my data to an AI company?

Accepted Answer

No. The platform calls no outside AI service, and FinBERT runs inside the platform's own services. The privacy page lists every third party that receives anything: your sign-in provider; Google as host, mail carrier and, if you open it, booking calendar; consent-based Google Analytics on public pages; Yahoo, which sees the ticker you open but not who asked; and any webhook you set up. None of those roles is an AI service.

Question 22

Are the forecasts always right?

Accepted Answer

No. They are estimates, and estimates can be wrong. Forecasts are scored against what happened once they mature, and a model that fails its statistical gate carries no trading weight. The platform's own standard is that a live record of a few months shows the system operates as designed, not that it has a statistically significant edge. Nothing on the site is investment advice, and past performance does not guarantee future results.

Question 23

How are the Opulence Alpha studies measured?

Accepted Answer

Every study uses one universe and one method. It observes 1,767 US-listed common stocks across 11 sectors every Wednesday from January 4, 1995 to August 19, 2026: 1,634 Wednesdays spanning 7,959 trading sessions. Each stock's return to the close 1, 5, 21 and 63 trading sessions later (roughly a day, a week, a month and a quarter) is compared with the same Wednesday's median stock. A relation is called proven only when its t-statistic on non-overlapping dates reaches 3 and it holds in most years.

Question 24

What does "no proven edge" mean in the studies?

Accepted Answer

It means the signal was measured and did not clear the bar for a proven relation. The methodology's verdict table splits that into two labels: "no measurable edge" means |t| below 2, and a reading of |t| 2 or more that still falls short of the proven bar is "faint". The scoreboard's "No proven edge" covers both, so a few of those beliefs, such as the MACD crossover, show a faint lag on their pages. It never means the idea went untested.

Question 25

Why compare every stock with the same day's median stock?

Accepted Answer

Because half of all stocks beat the median by construction, chance is 50 in 100 on every date, whatever the market did. A market-wide rally therefore cannot flatter a signal's result. The question becomes: did stocks showing this signal do better than the typical stock that same week? Average excess returns use the same day's equal-weighted average instead, and rank correlations are unaffected by either choice.

Question 26

What did the 24 popular trading beliefs show?

Accepted Answer

None showed a proven edge in the direction traders believe. Of 24 beliefs, 20 fell short of a proven edge and are marked "No edge", and 4 went measurably the other way: the rounding bottom, breakouts to a new 10-day high, Bollinger band breakouts and the stocks analysts cover most. The 22 headline rates ran from 48 to 51 in 100 against a coin-flip 50. Stocks on the day of a golden cross, for example, beat the same day's median stock over the next month 50 times in 100.

Question 27

What does "went the other way" mean?

Accepted Answer

It means the data showed a proven relation, but opposite to the belief. Stocks closing above their 10-day high are supposed to keep running; instead they beat the same day's median stock over the next week 49 times in 100, and lagged in 27 of 32 years. The shortfall is small but clears the proven bar. It is a tendency across hundreds of stocks, not a forecast for any single one.

Question 28

What counts as "proven" in the studies?

Accepted Answer

Three conditions must hold together. The t-statistic on non-overlapping dates must reach 3 in absolute value; the yearly average must point the same way in at least 60% of calendar years; and the effect must be large enough to matter, meaning a rank correlation of at least 0.01 or a hit rate at least 1 point from 50. The |t| ≥ 3 bar follows Harvey, Liu and Zhu (2016).

Question 29

How do the studies avoid results that are just luck from testing many ideas?

Accepted Answer

By counting every test and setting a high bar. Across all studies, 1,470 t-statistics were computed. If no signal worked, about 4 would reach |t| ≥ 3 by luck alone; 176 did. As a cross-check, a Benjamini–Hochberg control of the false discovery rate at 5% keeps 190 results, and every result called proven clears it. Many tests are correlated, so these counts are a guide rather than an exact rate.

Question 30

Does anything in the studies show an edge?

Accepted Answer

Yes. The 24 belief pages test popular claims one by one; signals that clear the bar are reported in the research papers, and five of the seven report a proven relation. In short-term reversal, for example, the fifth of stocks that beat their industry most over three sessions beat the median stock the next week 48 times in 100, the same way in 32 of 32 years. These are small tilts, not forecasts.

Question 31

How do the studies deal with survivorship bias?

Accepted Answer

The universe includes every company in the S&P 500 at any time since 1996 whose price history could be recovered: 1,027 companies, 413 of which have since delisted and stay in with full history. The 137 missing past members mostly left the index before 2010. The 740 companies outside the index were chosen from those listed when the universe was built, so smaller companies that failed or were acquired earlier are under-represented. Each lead result is therefore re-measured on the stocks that were S&P 500 members on each date.

Question 32

Why do the studies sample stocks every Wednesday?

Accepted Answer

The studies take one observation a week, and Wednesday keeps it away from the start and the end of the trading week. Weeks whose Wednesday was a market holiday, 17 of 1,651, are left out. The signal studies hold 1,889,758 stock-weeks from 1,763 companies, a median of 1,148 stocks per Wednesday. The chart-pattern studies also leave out weeks the regime model flags as a likely change of market regime: 101 of 1,651 Wednesdays, holidays included.

Question 33

What are the limitations of the studies?

Accepted Answer

The methodology states them plainly. Results are before trading costs, taxes and market impact. Returns start at the same close the signal is computed from, with no delay. No delisting return is added. The smaller companies in the universe are survivors. Every stock counts equally, so results describe the typical stock, not a capitalisation-weighted portfolio. And signals are recomputed over the past with today's definitions, so a relation found is a measured tendency, not the record of a strategy run live.

Question 34

Can I check the study numbers myself?

Accepted Answer

Yes, at the level of the published results. Every figure on the study pages is quoted from one frozen evidence snapshot, published free as JSON at /api/public/evidence with no sign-in. It holds the sample description, the per-signal results, the robustness checks and the test counts, and the same snapshot feeds the research console. It covers universe-wide studies only: no stock-level data is published, so you can check the figures but not re-run the studies from raw prices.

Question 35

How current are the study figures, and when are they updated?

Accepted Answer

In the current snapshot the studies were run on September 25, 2026, and their outcomes reach August 21, 2026. Figures are frozen in that snapshot and the study pages quote it and nothing else, so no sentence carries a hand-typed number. When the underlying data is corrected, the studies are re-run and the snapshot is re-dated. The date on every study page shows which vintage it quotes.

Question 36

Are the studies investment advice or stock forecasts?

Accepted Answer

No. The studies describe tendencies across hundreds of stocks. None of them forecasts a single stock's price, and none is investment advice. A measured tendency, even a proven one, is only a nudge for any one stock, and the results are before trading costs. Opulence Alpha publishes them as research, not as recommendations.

Question 37

What are Golden Tickets?

Accepted Answer

Golden Tickets are Opulence Alpha's weekly benchmark engine: a 10-stock basket that measures what the forecasts alone are worth, before any portfolio is built around them. They carry no costs, no risk overlay and no stops, and they are not a product an account can hold. Results are always reported in percent of the basket, never in currency.

Question 38

How is a Golden Ticket basket picked?

Accepted Answer

All four forecast horizons (1, 5, 21 and 63 trading days) blend into one conviction score at 22.4% / 15.8% / 24.8% / 37.1%. The platform's search engine found those weights; they were not set by hand. The top 10 names by that score form the basket, a name with a negative forecast enters short, and the basket leans however the cross-section leans. It is re-picked every week and held for the trading week, then the next basket is built from scratch: a name that ranks again is a new position, not a held one.

Question 39

Should I trade the Golden Tickets?

Accepted Answer

No. The Golden Tickets are a benchmark, not a recommendation. The basket is picked by rule and followed exactly as picked: every week is published, whether it rose or fell, so the forecasts can be judged on all of their results. It is a research record, not advice to buy or sell any stock.

Question 40

What are the Equal, Conviction and Kelly baskets?

Accepted Answer

All three hold the same 10 names each week and differ only in how the basket is split between them, so the spread between them is the price of the sizing decision alone. Equal gives every name the same share. Conviction raises a name's share with the strength of its forecast. Kelly scales the share with edge against uncertainty and takes 50% of the full Kelly bet; because the basket is always scaled to the same total, Kelly changes the shape of the bet across the names, not how much is at risk.

Question 41

How has the Kelly basket done?

Accepted Answer

Through the week that closed on 21 September 2026, the Kelly basket was down 12.10% and the Equal basket, holding the same names, was up 29.17%, both gross of costs. The baskets are followed exactly as picked, and every week of all three stands on the ledger as published.

Question 42

When are the Golden Ticket picks made and published?

Accepted Answer

Each basket is picked after Monday's close (the next trading day's when Monday is a holiday) and held to the following week's roll. Signed-in accounts, including the free 30-day trial, see the names on Monday night; the public ledger shows the basket's shape until the week closes, then every name, side and price.

Question 43

Do Golden Tickets have sector or position limits?

Accepted Answer

No, deliberately. The basket has no position caps, no sector or industry limits, no regime filter, no persistence, no cost model, no exit stack and no risk gate. Each of those is portfolio construction, and adding any of them would move the number away from the forecasts the benchmark exists to isolate.

Question 44

Are the returns net of trading costs?

Accepted Answer

Not for the Golden Tickets: no commission, slippage or borrow cost is charged, so the published number is the forecasts' yield and nothing else. The Growth, Moderate and Conservative books are charged a modelled cost on every fill: each stock's own measured bid-ask spread (capped at 60 basis points), a 0.5-basis-point commission, and a market-impact charge on orders large enough to move the price. In both cases the S&P 500 comparison is price return (SPY), so index dividends are left out.

Question 45

What are the Growth, Moderate and Conservative books?

Accepted Answer

They are the platform's three model portfolios, one for each risk profile the sign-up questionnaire can assign. All three size positions by the model's conviction and blend the 5-, 21- and 63-day forecasts equally; they differ in how much risk they may take. Conservative has the lowest ceiling on any one position and Growth the highest, and Growth alone targets gross exposure above 100%. The distance between a book and a Golden Ticket is therefore the cost, or the gain, of how the book is built: its horizon weighting, sizing, limits and risk rules.

Question 46

Does Opulence Alpha manage real money?

Accepted Answer

No. Opulence Alpha is research software: it does not hold client money, does not connect to a broker and does not place orders. The Golden Tickets are a weekly benchmark, reported in percent of the basket. The Growth, Moderate and Conservative books are the platform's model portfolios, run every trading day at each risk profile. Acting on any of it is your decision, placed through your own broker.

Question 47

How is fair value calculated?

Accepted Answer

Fair value is a model estimate. For most companies we project cash flows from the latest annual report and discount them at a CAPM-based cost of capital: a 5.0% equity risk premium, a risk-free rate mapped from the Fed-policy regime reading (2.0% to 5.5%), credit spreads and the stock's beta. We run 10,000 scenarios, each with its own growth and discount rates, and blend the result with the company's own earnings and EBITDA multiples and book value. Banks, insurers, REITs, utilities and pre-revenue healthcare companies get industry-specific models.

Question 48

Why are some companies set aside from the fair-value figures?

Accepted Answer

Two groups are left out of the figures and lists. Names whose cash-flow value fell below the three-month model's price range, the most bearish readings, are set aside because the service swaps that value for the model's median price, a price forecast rather than a fair value. So are names the model flags as implausible, such as a fair value above three times the price. On Sep 25, 2026 that was 382 and 47 of 1,227 valued names, so the headline figures exclude the most bearish readings.

Question 49

Is fair value a price target?

Accepted Answer

No. Fair value is a model's estimate of what a company's cash flows, multiples and book value imply it is worth. It is a long-run value estimate, not a forecast of where the price will trade on any date, and not a recommendation to buy or sell. A stock can trade below its estimated value for a long time. Beside each gap the page shows P(under), a separate model's probability of beating the average stock over three months.

Question 50

How often is fair value updated?

Accepted Answer

After every US trading day. The valuation service recomputes each covered company's fair value with that day's close and the latest discount-rate and market-regime inputs, and the page states which close it reflects. Cash flows are projected from the company's latest annual report, so that part changes when a new report is filed. The public page shows the market-wide picture and the ten names furthest either side; every covered name, every day, is in the console.

Question 51

Which companies get a fair value?

Accepted Answer

We value the US companies in our coverage that are listed today: 1,266 on the Sep 25, 2026 close, of which 1,227 had a fair value and 39 did not. The public figures use the names that are not set aside, and the top-ten lists also require a market capitalisation above $1 billion. Companies that have since delisted stay in our studies with their full history, to limit survivorship bias, but no longer trade and get no fair value.

Question 52

Does a large gap mean the price will rise soon?

Accepted Answer

Not necessarily. The gap compares a long-run value estimate with the last close and says nothing about timing. P(under) is a separate model's three-month read, so it can differ from what the gap alone suggests; read the two together.

Question 53

Does the market regime change fair value?

Accepted Answer

Yes. The fair-value model reads the same regime record as the Regime Radar every day. The monetary-policy reading sets the risk-free rate on a scale from 2.0% to 5.5%. The composite regime lowers the ceiling on long-run growth and adds to the discount rate in risk-off markets, and trims the discount rate slightly in a strong risk-on market. A change of regime can therefore move many fair values at once.

Question 54

What is the Regime Radar?

Accepted Answer

The Regime Radar is Opulence Alpha's free daily reading of the US market regime. It models nine structural axes, from volatility and credit to inflation and monetary policy, each on its own, and fuses them into one composite regime on a six-step scale from strong risk on to crisis. It shows how long the regime has held, the model's chance that it changes, and how US sectors behaved in past sessions with the same regime.

Question 55

What are the Regime Radar's nine axes?

Accepted Answer

Volatility structure, market sentiment, market liquidity, credit conditions, recession risk, the fundamental cycle, the inflation environment, real interest rates and monetary policy. Each axis has its own indicators and named states, for example credit conditions from benign to distressed and monetary policy from very accommodative to very restrictive. For each axis the page shows today's state, a confidence, and a sector tilt: each GICS sector's historical sensitivity to that axis, in basis points.

Question 56

How often does the Regime Radar update?

Accepted Answer

Every US trading day. The nine axes and the composite are recomputed in the platform's daily run, and the page states the date of the reading it shows. Economic releases enter on the day they were published, so no reading uses data that was not yet out. Axes publish independently; if one has not updated, the page shows that axis's own date and marks it as behind the latest reading rather than presenting it as current.

Question 57

How is a market regime detected?

Accepted Answer

Each axis is read on its own: its indicators are placed in their own history, and percentile thresholds set for that axis decide its state and confidence. A hidden Markov model and a gradient-boosted classifier are trained per axis only as cross-checks; the published states come from the percentile classifier. The nine readings are fused by a hidden semi-Markov model, which models how long regimes tend to last and switches only when evidence for a new state clears a margin, so one noisy day does not flip the composite.

Question 58

Does the Regime Radar forecast prices?

Accepted Answer

No. Its forecasts are of the regime, never of a price: the chance the composite regime changes tomorrow, and the chance it changes within five sessions with the model's interval. The sector tables show how US sectors did against the S&P 500 in past sessions with the same regime, measured on next-session ETF price returns. They describe history, not what will happen, and are not investment advice.

Question 59

How far back does the regime history go?

Accepted Answer

The composite regime record starts on Jan 3, 1983. Sector comparisons, which use sector ETF price returns, start on Dec 22, 1998; funds that started later are measured from their first day. For today's composite the page shows the share of trading days it has held since 1983, the number of separate episodes, and each sector's annualised next-session excess return over the S&P 500 in those sessions.

Question 60

Do you offer access for universities and students?

Accepted Answer

Yes. Students of graduate courses in quantitative finance or data science can have six months of free access to the Opulence Alpha platform. Instructors and researchers can use the published studies and the dataset behind them for teaching and research, and we design joint research with academics aimed at publication. To arrange access for a course, write to us through the Contact page with the university, the course and the number of students.

Question 61

What can students do on the platform?

Accepted Answer

Students work on the same platform subscribers use, because every plan runs the same models and services. They can read forecasts at 1-, 5-, 21- and 63-day horizons with prediction intervals, fair-value estimates, market-regime readings and portfolio risk checks. They can also test a strategy through walk-forward validation, which fits only on what was knowable on each date, and follow it as a test portfolio that holds no money. The platform holds no assets, connects to no broker and places no orders.

Question 62

Can academics use the studies and data in their research?

Accepted Answer

Yes. Every study is free to read and carries a citation line and a link to the evidence snapshot, the JSON file behind every figure it quotes. The studies observe one universe of 1,767 US stocks every Wednesday since January 1995. It includes 413 companies that were in the S&P 500 at some point since 1996 and later delisted. That largely removes survivorship bias from the large-company layer, though smaller companies that failed or were acquired earlier remain under-represented. Our academic programme also opens the dataset behind the studies for teaching and research.

Question 63

How do I start a research collaboration?

Accepted Answer

Write to us through the Contact page, choosing General message, and describe the question you want to study, your institution and your timeline. A person reads and answers every message. Joint projects are designed together from the start, with academic publication as the aim, and can draw on the same universe, data and methods as the published studies. Opulence Alpha is based in Maslak, Sarıyer, Istanbul.

Question 64

Can I run out of credits mid-analysis?

Accepted Answer

No. Priced work is charged when it starts, never partway through, so no run stops halfway for lack of credits. Professional carries a 100-credit overdraft buffer, so a run may start even if it takes the balance up to 100 credits below zero; the next run is quoted against what is left. Investor has no buffer, so a run your balance cannot cover waits until you add a pack. Institutional's buffer is set by agreement. No pack is ever bought automatically.