He had a point: at picking stocks, skill and instinct rarely beat blind luck. We took it further: what if you replaced the monkey with a systematic process, applied it to thousands of companies, and then ran the results through every stress test we know?
That is exactly what Beat The Market does.
"A blindfolded monkey throwing darts at a newspaper's financial pages could select a portfolio that would do just as well as one carefully selected by experts." We took that seriously.
Burton G. Malkiel · A Random Walk Down Wall Street, 1973
Here is the exact sequence, applied to every strategy on the leaderboard.
Every strategy on this platform originates in published academic research on market anomalies, things financial economists have studied, debated, and replicated across decades of data. We do not start with a promising-looking chart and work backward. We start with a documented phenomenon and ask whether the data supports it on a fresh dataset.
The universe covers thousands of US-listed equities, sourced from SEC filings and market data, and updates regularly. The eligible pool is made up of US companies listed on the New York Stock Exchange and Nasdaq.
This is where most claims fall apart. A strategy can look great in a backtest simply because someone went looking for patterns until they found one. We run each candidate through up to 400 distinct configurations, then subject survivors to 28 combinatorially purged cross-validation paths designed to detect exactly that problem. Most candidates do not survive this stage. Walk-forward tested across 143 months of market history, nearly 12 years, including companies that later left the market during the period.
From the moment a strategy goes live, we track it daily. Returns, risk statistics, and rank are recalculated using real market data. The leaderboard shows the live record alongside the backtest history so you can see exactly how the strategy is performing against what was expected.
What you do with this information is entirely your decision. Beat The Market is an information publisher, not a registered investment adviser. Model allocations reflect the strategy's construction weighting. They are not instructions.
These are the numbers behind every single strategy we publish. Not the whole board, one strategy. Each one runs this gauntlet.
Each configuration is a full backtest spanning more than 12 years of history, re-ranking thousands of companies at every monthly rebalance. We run up to 400. One is selected and published.
Every strategy at Beat The Market traces to a published, peer-reviewed paper. We do not generate strategy ideas internally and then go looking for supporting research. The paper comes first. All six strategies currently live on the leaderboard ground on the following research.
Walk-forward testing. You split historical data into an earlier training window and a later testing window. The strategy is designed on the training data, then applied to the testing data without modification. If it holds up in the testing window, that is evidence the rule generalizes beyond the data it was calibrated on.
Probability of backtest overfitting (PBO). When you test many strategy variations and pick the best performer, you are likely to select one that looked good by luck rather than design. PBO methods quantify that risk by re-running the selection process across many data permutations and measuring how often the winning selection also leads in the later, held-out period it was not calibrated on.
Combinatorial purged cross-validation (CPCV). A more exhaustive cross-validation technique that generates many train-test splits from the same data set and evaluates strategy performance across all of them. Reduces the variance of the overfitting estimate.
Deflated Sharpe ratio (DSR) and probabilistic Sharpe ratio (PSR). Standard Sharpe ratios do not correct for the fact that you are selecting from a large set of trials. DSR and PSR adjust for the number of configurations tested and the statistical properties of the distribution, producing a more honest estimate of whether the observed ratio is likely to reflect a true signal rather than sampling noise.
See which strategies have passed the gauntlet, what they hold, and how they are performing since launch.