Random Entry Test: Is Your Strategy's Edge Real?
What if a coin flip could match your entry rule? A random entry test answers exactly that question. You keep your exits, fees and position size, then replace the entry with chance. If random entries do as well as your rule, the entry adds nothing.
We ran the test on a real strategy from this blog. The results were more mixed than the original backtest suggested.
What Is a Random Entry Test?
A random entry test is a backtest check that compares your strategy against the same strategy with random entries. Everything else stays fixed. Only the timing of each entry changes.
You run the random version hundreds or thousands of times. That gives you a spread of results that luck alone can produce. Then you see where your real result falls in that spread.
If your strategy beats 95% of random runs, chance is an unlikely explanation. If it beats 60%, a coin flip could easily have done the same.
Why a Backtest Alone Is Not Enough
A profitable backtest proves less than it seems. Several things can make money that have nothing to do with your entry rule.
The market went up. In a strong bull market, almost any long entry makes money. Bitcoin doubled over the period we tested.
The exits did the work. Our guide to entry and exit strategies covers how the two interact. A good stop and exit can turn random entries into a small profit. Or they can lose money on their own. Either way, you need to know.
Luck. With 60 trades, a handful of big winners can swing the total. A statistical significance check helps, but the random test shows the spread directly.
How to Run a Random Entry Test
The test has four steps.
- Freeze the rest of the strategy. Keep the same exits, stop, fees, market, timeframe and dates.
- Match the trade count. Set the random entry chance so the random version trades about as often as yours. A rule that trades less pays fewer fees, which is not an edge.
- Run it many times. We used 1,000 runs per market. Fewer than a few hundred gives a rough picture.
- Rank your result. Count the share of random runs your strategy beat.
What to Hold Constant
Long-only rules should get long-only random entries. If your rule has a trend filter, apply the same filter to the random entries. Otherwise you are testing the filter and the entry together.
What Happened When We Tested a Real Strategy?
We used the volume oscillator strategy published yesterday. It buys a 20-candle breakout only when volume is above normal. It exits on a 10-candle low or a stop two Average True Ranges (ATR) below entry. ATR measures typical candle movement.
The data was Binance four-hour candles from 10 November 2023 to 10 November 2026. Fees were 0.1% per side.
| Market | Strategy return | Strategy trades | Random median | Random runs in profit | Strategy beats |
|---|---|---|---|---|---|
| BTC | +46.5% | 61 | -13.3% | 30% | 97% of runs |
| ETH | +66.1% | 67 | -9.0% | 40% | 94% of runs |
| SOL | +18.6% | 75 | -28.2% | 24% | 84% of runs |
Hypothetical rule tests on past data, with no slippage. They are not trading results, and past behaviour may not repeat.
Reading the Table
The first finding is about the exits. The median random run lost money on all three markets. That happened even though BTC rose 105% and SOL rose 83% over the period. The exits and fees lose money on their own, so any profit must come from the entries.
The second finding is about the entries. The strategy clears the common 95% bar on BTC. ETH is just under it. SOL, at 84%, is the kind of result chance produces about one time in six.
The Plain Breakout Did Worse
We also tested the plain breakout, with no volume filter. It traded about 104 times per market. We matched the random runs to that count.
It beat 78% of random runs on BTC, 56% on ETH and 82% on SOL. On ETH, the plain breakout was barely better than a coin flip. The volume filter is what moved the strategy toward the top of the range.
Where Random Entry Tests Mislead
One market is not enough. A rule that beats 97% of random runs on one market may be lucky there. Test several markets.
The random runs are not your future. The test tells you how unusual your result was on past data. It does not promise the edge will last.
Testing many rules breaks the bar. If you test 20 rules, one will probably beat 95% of random runs by luck. Our post on the deflated Sharpe ratio explains how to adjust for that.
How to Apply Random Entry Tests in Arrow Algo
Arrow Algo strategies are built from visual blocks, with no code. A random entry version needs only the entry swapped out.
- Duplicate your finished strategy.
- Delete the entry condition blocks, but keep the flat-position check and every exit block.
- Add a random number block that outputs a value between 0 and 1 on each candle. Connect it to a condition block that checks it is below a small threshold.
- Adjust the threshold until the random version trades about as often as your real rule.
- Run the duplicate many times on the same data and fee settings, and record each result.
Running hundreds of versions by hand is slow. A connected AI assistant, such as Claude or ChatGPT through the Arrow Algo connector, can run and log the batch for you. Check the block graph of the first run to confirm it changed only the entry.
Nothing needs to go live for any of this. If you later choose to deploy, your rules signal your own exchange account. The exchange decides whether to execute each order.
What Are the Key Takeaways?
- A random entry test keeps your exits and costs, and replaces only the entry with chance.
- Match the random trade count to your strategy, or fees will distort the comparison.
- In our test, the exits alone lost money, even in rising markets.
- The volume filter strategy beat 97%, 94% and 84% of random runs on BTC, ETH and SOL.
- The plain breakout beat only 56% of random runs on ETH, barely better than chance.
- Run the test on several markets, and adjust for how many rules you have tried.
Disclaimer: This content is for educational purposes only and does not constitute financial advice. Trading involves significant risk and you should only trade with capital you can afford to lose. Past performance is not indicative of future results. Always conduct your own research before making any trading decisions.
Ready to build your own automated trading strategies without writing a single line of code? Start for free at Arrow Algo and join thousands of traders who've made the switch to systematic trading.