RANDOM HARVESTERHistorical portfolio research for thoughtful investors
See how Random Harvester would have handled your assets.
Choose the stocks or ETFs you want to examine. See how a rules-based historical simulation would have traded them, what happened to gains moved out of the trading account, and how total wealth compared with holding your assets.
Research preview · No payment today · No live trading
Cash and holdings still exposed to the strategy.
Gains moved outside the simulated trading account.
The two amounts together, compared with alternatives.
Concept illustration; no performance figures are shown. Open the example PDF for simulated results.
How it works
Start with your own selection.
Choose assets
Enter 5 to 10 stock or ETF tickers and one total starting amount. The amount is divided equally across the assets.
Choose a period
Select the current year, the last five years, or your own historical dates.
Inspect the method
Open the free report example to see the charts, comparisons, assumptions, and limitations a full report can contain.
What a report can answer
Understand the journey, not just the final number.
The report is designed to show where the approach helped, where it struggled, and how much capital remained exposed.
When would the rules have traded?
Review simulated entries, exits, and time outside the market for your chosen assets.
What did banking gains change?
Compare total investor wealth with and without profit skimming under the same trading rules. Either outcome can finish ahead.
Which assets drove the outcome?
See results per asset, including losses and periods of difficult trading.
How did alternatives fare?
Compare with passive holding of your assets and relevant benchmarks using return and risk measures.
Backtesting discipline
How we address common backtesting pitfalls.
A historical simulation is only useful when its assumptions and limitations are visible. Random Harvester is designed to reduce four common sources of misleading results.
The pitfall
Parameter overfitting
Trying enough parameter combinations can produce a persuasive result by chance.
Random Harvester’s response
The configuration is fixed before each declared test and is not re-tuned as the test progresses. This reduces parameter-search bias, but it does not eliminate overfitting; historical results remain in-sample until confirmed on untouched data or through live forward testing.
The pitfall
Regime dependence
A strategy tested over one short or favourable period may only fit that market.
Random Harvester’s response
The same configuration is examined across distinct regimes, including the 2007-09 systemic crisis, the 2020 crash, the 2022 bear market, and calmer recovery periods. Results reported by era, including failures, are available.
The pitfall
Unrealistic execution and costs
Backtests can assume instant fills at known prices and omit trading frictions.
Random Harvester’s response
A decision made at time t is scheduled for the next available bar. The simulation applies configured broker fees and transaction taxes to trades and accounts separately for capital-gains tax on banked profit. It does not model live slippage or liquidity effects.
The pitfall
Look-ahead bias
Future observations can leak into a signal through centred averages, revised indicators, or other retrospective calculations.
Random Harvester’s response
Within the prepared price series, the detector processes observations from left to right. A decision at time t uses only information available through time t; it does not use future bars, centred averages, or rewrite earlier signals. Vendor-adjusted histories can still be revised after corporate actions.
Free example
See the report before you decide.
The EuroStocks example shows historical simulations, banked profit, asset analysis, and benchmark comparisons. It includes 14 assets as an example of report format; requests on this page accept 5 to 10.
Random Harvester tests each asset using its own historical price data. It looks for sustained trends and checks their consistency before simulating trades. Results for individual assets show whether a portfolio outcome depends on just a few holdings.
After a profitable sale, the simulation can move part of the gain out of the trading account as banked profit. It is still part of total investor wealth, but it is no longer exposed to later trades in the strategy. With less capital available for those trades, banking can help or hurt the final result depending on what happens next.
Open the example PDFClear assumptions.
15-page example report
Simulated results · September 2026
Interest survey
Which portfolio would you test?
Tell us which assets you would examine and what you most want to understand: trading decisions, banked gains, asset-level outcomes, or benchmark comparisons. We are gathering requests to shape the service. Submitting this form does not place an order or produce a report.
We will use your email only to follow up about this request. Please read our privacy policy.
A clear view of the limits.
Random Harvester tests historical data with a causal, long-only simulation and compares results with reference strategies. Backtests depend on the chosen assets, data, assumptions, and market conditions. The current results are in-sample: parameters were calibrated with knowledge of the evaluation data. Past simulated results do not predict future returns. This service does not execute trades or provide personal investment advice.