Education
Our 2026 Backtest Results Explained: Compounding, Fixed Lot and the Deepest Baskets
SmartEA Research, 2026-10-28, updated 2026-11-19, 8 min read

Short answer: we test every EA twice. Once with the lot growing with the balance (our default, 0.01 lot per $1,000), which shows what the EA does when profits are reinvested, and once with a fixed 0.01 lot, which shows the strategy itself without compounding. Both point to the same risk: the deepest equity drop in 2026 was about 21% of the account for Smart Rider, 14 to 17% for Smart Striker, 27% for Smart Archer and 6 to 15% for Smart Breaker BTC. For Rider, Striker and Archer each of those drops was a floating loss in an open basket or open positions; in the grid EAs no deep basket closed with a loss. Striker closes losing trades with a stop loss.
On this page you find both sets of results for all four EAs, how to read them, and the deepest baskets of the year with dates, so you can look them up in your own backtest.
How we test
- Period: 1 January to 26 October 2026 (Smart Breaker BTC: 1 January to 18 November 2026, after its six new strategies)
- Symbol: XAUUSD on H1, every tick based on real ticks, broker server time GMT+2 / GMT+3
- Account: $1,000 start, leverage 1:500, 300 ms execution delay
- Settings: the EA's defaults, all strategies on, drawdown protection off
You can repeat every test yourself for free: how to backtest SmartEA gold EAs.
Why two tests
With our default risk the EA sizes each new basket from the current balance: 0.01 lot per $1,000. When the balance doubles, the lot doubles. Over ten good months this compounding turns $1,000 into a very large number. That number is correct for the test, but it says more about compounding than about the strategy, and it hides that losses in money grow with the lot too. To play with the arithmetic of compounding yourself, try our forex compounding calculator.
The fixed-lot test keeps 0.01 lot all year. Profit grows in a straight line instead of a curve, and every number is in money per 0.01 lot: the size of the worst floating loss, the profit of a normal month. That makes the risk easy to translate to your account: at 0.01 lot per $1,000, a $229 drop per 0.01 lot is about 23% of the account, whatever its size.
The results
Smart Rider Gold EA
- Lot grows with the balance: $1,000 to $70,405, max drawdown 20.9%, profit factor 5.04, 905 trades.
- Fixed 0.01 lot: profit $4,516 (end balance $5,516), max drawdown $229, profit factor 7.13.
- Baskets: 527 in the year, none closed with a loss.
- Version: these are version 1.01 figures (November 2026). Its new entry filters make it trade less in 2026 than version 1.00, but every year from 2020 to 2026 ended in profit; the year-by-year figures are on the EA page.
Smart Striker Gold EA
- Lot grows with the balance: $1,000 to $11,230, max drawdown 14.2%, profit factor 4.17, 533 trades.
- Fixed 0.01 lot: profit $3,317 (end balance $4,317), max drawdown $170, profit factor 6.06.
- Losses: Striker is not a grid EA: every trade has its own stop loss, so it closes losing trades (about a quarter of its trades). The longest losing streak was 9 trades in a row in early October, together $50 per 0.01 lot.
- Weekend: since November 2026 Striker removes its waiting stop orders from 23:00 server time on Friday until Monday (on by default), so a weekend gap cannot open a trade; these figures include that.
Smart Archer Gold EA
- Lot grows with the balance: $1,000 to $127,455, max drawdown 26.7%, profit factor 3.19, 2,427 trades.
- Fixed 0.01 lot: profit $5,451 (end balance $6,451), max drawdown $267, profit factor 3.64.
- Baskets: 1,300 in the year; 4 closed slightly below zero (between $0.35 and $4.84 per 0.01 lot), all others in profit.
Smart Breaker BTC EA
- Period: 1 January to 18 November 2026, version with 19 strategies.
- Lot grows with the balance: $1,000 to $7,544, max drawdown 6.3%, profit factor 2.19, 10,577 trades.
- Fixed 0.01 lot: profit $3,577 (end balance $4,577), max drawdown $150, profit factor 2.01.
- Losses: Breaker trades Bitcoin (BTCUSD), not gold, and every trade has a stop loss. About six of seven trades close in profit, but the average loss is about three times the average win. The longest losing streak was 14 trades in a row on 2 February, within four minutes, together $44 per 0.01 lot.
Two things stand out. First, the drawdown is about the same in both tests: 21% with compounding against $229 on a $1,000 start for Rider, 27% against $267 for Archer. That is not a coincidence: with the lot tied to the balance, a basket that costs $229 per 0.01 lot costs the same share of any account. Second, the profit factor is higher with a fixed lot, because with compounding the later, larger trades count more in the totals.
The deepest baskets of 2026
A grid EA has no stop loss per trade. When price moves against a basket, it adds levels with the same lot and waits for price to come back to a common target. While it waits, the basket shows a floating loss: a loss on paper that becomes real only if the basket is closed at that moment. These are the baskets with the largest floating loss in 2026, measured on every tick per 0.01 lot level (so a figure can be a little higher than the maximum drawdown in the tester report):
Smart Rider Gold EA
- 1 May, S04 (buy), 6 levels: worst floating loss about $229 per 0.01 lot, after about 5.5 hours. Closed at its target the same afternoon with +$32. In the compounding test this was the deepest point of the year: 20.9% of the balance at that moment.
- 17 February, S03 (sell), 5 levels: about $187 at the worst point, closed after 7.7 hours with +$44.
- 23 June, S03 (sell), 5 levels: about $187, held for 14 hours over the night, closed with +$9.
Smart Archer Gold EA
- 20 March, S15 (sell), 10 levels: about $300 per 0.01 lot at the worst tick, a fast spike up that reversed within half an hour; closed with +$46.
- 26 October, S14 (sell), 14 levels: about $277, closed after 5 hours with +$11. With compounding this was the deepest point of the year, 26.7%, because the balance and the lot were largest by then.
- 29 April, S14 (sell), 13 levels: about $276, closed after 3.6 hours with +$14.
None of these baskets lost money: each one came back to its target. But at the worst moment each of them was a real risk. If gold had kept moving in the same direction, the floating loss would have kept growing, because a grid has no maximum number of levels. That is the price of the high win rate, and it is why we show these numbers instead of only the final balance.
What this means for your settings
- Plan for the worst basket, not the average month. With the default 0.01 lot per $1,000, expect drops like the ones above (about 21% for Rider, 27% for Archer, 14 to 17% for Striker, 6 to 15% for Breaker), and larger ones if the market does something it did not do in 2026.
- Lower the risk with the balance per 0.01 lot. At 0.01 lot per $2,000 every drop above is roughly halved, and so is the profit.
- Use the drawdown protection at a level you can accept. It turns an open-ended worst case into a known maximum loss, at the cost of closing some baskets that would have recovered.
- Small accounts: below $1,000 use a cent account, so the lot can follow the balance. Our cent vs standard account backtests show how much faster the lot compounds there.
What a backtest does not show
These are tests on one broker's past prices. Live trading adds slippage, wider spreads around news and rollover, and markets that were not in the test. Compare the tests with the live accounts linked on each EA page and with our Telegram channel, where the result of every closed basket is posted, losses included. For how to read a tester report in general, see how to read EA performance statistics.
Comparing our results with other gold EAs? Use the same checks we describe in how to choose a gold EA for MT5: equity drawdown, the deepest losing basket and how the EA limits losses.
Frequently asked questions
Why is the compounding end balance so high?
Because every basket is sized from the balance at that moment, and ten profitable months compound. The fixed-lot test shows the same strategy without that effect: about $4,500 per 0.01 lot for Rider, $5,500 for Archer, $3,300 for Striker and $3,600 for Breaker (to 18 November).
Which max drawdown should I look at?
The percentage in the compounding test, or the money figure per 0.01 lot in the fixed-lot test. Both describe the same drop. Ignore any drawdown percentage measured against a balance that has already grown a lot: it makes late drops look small.
Did any basket lose money?
In 2026 no Rider basket closed with a loss, and Archer had 4 baskets of 1,300 that closed a few dollars below zero. Striker closes losing trades by design, with a stop loss on each trade. A basket that has not lost money in a backtest can still lose money live, which is why we recommend the drawdown protection.
Backtests are simulations on past prices and do not predict future results. A grid EA has no stop loss per trade and no maximum number of levels. Test on a demo account first, use a risk level you can afford, and only trade money you can afford to lose.
Follow SmartEA Insider on Telegram
SmartEA Insider: results of our own live accounts, new versions and release notes. No trade signals.


