Articles · Mike Harty · Published 2026-10-24 · Updated 2026-10-24
How Much Capital Do You Need to Run an EA? Minimum Balance
Short answer: More than the vendor's "minimum deposit". Your real minimum is the larger of four numbers: the balance where 0.01 lot still matches your risk setting, the balance that survives a realistic losing streak, the balance that survives a tail drawdown (not the backtest's), and the balance with enough margin at full exposure. For a typical 1%-risk EA that's usually thousands, not hundreds.
Key takeaways
- A "$100 minimum" only means the EA can open a trade. It says nothing about whether your account survives.
- Below the 0.01-lot floor, your real risk per trade is higher than the EA's risk setting.
- Even an EA with a real edge hits deep drawdowns at 5% risk per trade. Size from the bad case, not the average.
- Pick your stop-loss in dollars before you go live, then work backwards to the deposit.
Why the vendor's minimum deposit misleads
Most EA product pages quote the smallest balance the broker will accept for the EA's smallest trade. That's a technical minimum. The number you need is the survival minimum: the balance that can absorb the strategy's normal bad patches without you panicking, the broker stopping you out, or the EA quietly running at 3× the risk you set.
Four constraints decide it. Work out each one, then take the biggest.
All numbers below are illustrative examples from stated assumptions: a EURUSD-style symbol, $10 per pip per 1.00 lot, 0.01 minimum lot and lot step, lots rounded down. They are not backtests or performance claims.
Check 1: The 0.01-lot floor
The EA sizes trades like this:
Lots = (Equity × Risk%) / (StopLoss_pips × PipValue_per_lot)
The broker won't go below 0.01 lot. So the smallest account where your risk setting is honest is:
FloorEquity = (0.01 × StopLoss_pips × PipValue_per_lot) / Risk%
| Stop loss | 0.5% risk | 1% risk | 2% risk |
|---|---|---|---|
| 20 pips | $400 | $200 | $100 |
| 30 pips | $600 | $300 | $150 |
| 50 pips | $1,000 | $500 | $250 |
| 100 pips | $2,000 | $1,000 | $500 |
What it means: run a 100-pip-stop EA on $200 and every 0.01-lot loss costs $10, which is 5% of the account. It doesn't matter that the input says "Risk = 1%".
Multiple positions (and multiple EAs) multiply the floor
Most EAs don't hold just one trade. If yours can hold N positions at once (several pairs, several magic numbers, or a basket), your worst-case open risk is N × per-trade risk. The same applies when several EAs share one account; see running multiple EAs on one MT5 account .
Illustrative: 30-pip stops, 1% each, up to 3 open trades. The floor is $300 per trade, but with all three open you have 3% of the account at risk. If you want total open risk capped at 1%, the floor becomes $900.
Small accounts have a second problem: no granularity. On $500, a single 0.01-lot step on a 20–50 pip stop is worth 0.4–1% of the account per trade, so sizing can't adjust smoothly as equity moves.
Check 2: Losing streaks are longer than you think
With fixed risk r per trade, k losses in a row take you down 1 − (1 − r)^k:
| Risk/trade | 5 losses | 10 losses | 15 losses | 20 losses |
|---|---|---|---|---|
| 1% | 4.9% | 9.6% | 14.0% | 18.2% |
| 2% | 9.6% | 18.3% | 26.1% | 33.2% |
| 5% | 22.6% | 40.1% | 53.7% | 64.2% |
How likely are these streaks? Here is the chance of at least one streak of length k or more within 500 trades (exact calculation):
| Streak | 45% win rate | 60% win rate |
|---|---|---|
| 6 | 99.9% | 71.0% |
| 8 | 85.4% | 17.7% |
| 10 | 43.3% | 3.0% |
| 12 | 15.6% | 0.5% |
A 45%-win-rate trend EA will almost certainly see 8 straight losses over 500 trades. That isn't the EA "breaking". It's normal operation, and your capital has to treat it that way.
And the climb back is steeper than the fall:
| Drawdown | Gain needed to recover |
|---|---|
| 10% | 11.1% |
| 20% | 25.0% |
| 30% | 42.9% |
| 50% | 100% |
Check 3: Size from the tail drawdown, not the backtest
Want to run this on your own backtest? Reshuffle its trades (Monte Carlo) to see a range of drawdowns, as VRD Quant does.
A backtest is one path. The future is a different one. To see the range, we simulated an EA with a genuine edge (45% win rate, 1.5:1 reward:risk, 1,000 trades, 20,000 runs). "Ruin" here means a 50% drawdown, the point where most people switch off:
| Risk/trade | Chance of a 50% DD | Typical (median) max DD | Bad-case (95th pct) max DD |
|---|---|---|---|
| 1% | 0.01% | 18.8% | 30.0% |
| 2% | 7.6% | 35.0% | 52.6% |
| 5% | 97.5% | 69.6% | 88.1% |
| 10% | 100% | 94.7% | 99.6% |
Two lessons: 1. A real edge doesn't protect you from oversized risk. At 5% per trade, this profitable system almost always halves the account at some point. 2. Risk doesn't scale linearly. Going from 1% to 2% takes the 50%-drawdown odds from about 1 in 10,000 to about 1 in 13.
Turn it into dollars
RequiredCapital = MaxLossYouWillAccept_$ / BadCase_MaxDD%
Illustrative: you'd stop the EA after losing $1,500. At 1% risk (bad case ≈30%) you need about $5,000. At 2% risk (≈52.6%) about $2,850, but you've signed up for roughly a 1-in-13 chance of a 50% drawdown.
If a vendor's backtest shows 15% max DD, read it as "at least 15%", never "at most".
Check 4: Margin and leverage
Your stop loss limits the loss. Margin decides whether the trade can exist at all, and whether the broker closes you out before your stop is hit.
Margin = Lots × ContractSize × Price / Leverage
EURUSD at 1.0800, 0.01 lot ≈ $1,080 notional:
| Leverage | Margin per 0.01 lot | Per 0.10 lot |
|---|---|---|
| 1:30 | $36.00 | $360.00 |
| 1:100 | $10.80 | $108.00 |
| 1:500 | $2.16 | $21.60 |
Grid and martingale EAs: a special case
Deep dive: Grid and martingale EAs: why smooth equity curves fail . Grid, averaging and martingale EAs usually have no fixed per-trade stop. That breaks Checks 1–3 because there is no "risk per trade" to plug in. Size them on the full basket:
BasketExposure_$per_pip = TotalOpenLots × PipValue_per_lot
PipsToWipeout = Equity / BasketExposure_$per_pip
Illustrative: 8 positions × 0.05 lot = 0.40 lot means $4 per pip. On $500, a 125-pip move against the basket wipes the account. At 1:30 leverage the basket needs $1,440 of margin, so it can't even open. At 1:500 it needs $86.40 and opens easily. High leverage doesn't make a grid safer. It removes the brake.
If an EA's most popular signal survived only because money was deposited during drawdowns, your capital has to include those deposits too. That's one reason our EA review framework penalises it heavily.
Worked example: a popular MQL5 recovery EA at 0.85 lots (our backtests)
These are real numbers from two MT5 Strategy Tester backtests we ran on a popular MQL5 recovery EA: GBPUSD H1, 1 Jan 2020 to 22 Oct 2026, 100% history quality, IC Markets data, $10,000, 1:500. They are backtests, not live results.
| 0.01 lots | 0.85 lots (5 × 0.17) | |
|---|---|---|
| Net profit | $817.78 | $80,187.21 |
| Profit factor | 1.73 | 2.39 |
| Max equity drawdown | $194.08 | $8,399.39 |
| Max balance drawdown | $115.99 | $5,062.47 |
| Worst month | n/a | −$3,958 (Apr 2022) |
Lesson 1: drawdown doesn't scale linearly with lot size. 85× the lot gave about 43× the drawdown. Scaling the 0.01 test up (×17 per leg) predicts a $3,322 worst drawdown at 0.85 lots. The real 0.85 run hit $8,399, about 2.5× more. At 0.01 lots the EA can't split trades into 5 legs or size up its recovery trades, so the small test is a different strategy. For recovery or multi-leg EAs, backtest at the lot you'll trade.
Lesson 2: operating capital and total commitment are different numbers. In a simulation that swept profits above a floor every Friday, 0.85 lots on a $4,000 floor never blew ($152/month over the last 12 months), and a $3,000 floor blew once (16 Oct 2025). That simulation used the linear scaling, though. The direct 0.85 run had a −$3,958 month that would have emptied a $4,000 floor. So:
| Passport line | 0.85 lots |
|---|---|
| Operating capital in the account | $4,000 |
| Central reserve allocation | ≈ $8,600 |
| Maximum commitment (1.5 × $8,399 worst equity DD) | ≈ $12,600 |
That is the idea behind the VRD Passport and central reserve in VRD Complete ($149). Each EA runs on a lean operating balance with its own limits. One shared reserve covers the refill after a bad run, so you don't park idle cash in every account. In our tests the vendor's default sizing guidance pointed to a much smaller deposit than the backtest drawdown implied, which is why we size from the backtest rather than the product page.
Caveats: - These are backtests only, on one broker's data. - A 5,000-run Monte Carlo and slippage stress (24 Oct 2026) used the same scaled model and gave a P95 drawdown of $2,337. That is a lower bound, well below the direct run's $8,399, so size from the direct run. No execution-delay test was run. See Monte Carlo for MT5 backtests. - The floor simulation scales the 0.01 run linearly ×17, which understated drawdown here. - The $4,000 floor depends on how the live lot opens. These tests assume 0.85 opens as 5 orders of 0.17. We haven't confirmed that the live account opens the same leg structure as the backtest. If it doesn't, the validated floor could be much higher (around $17,500 in our grid). - The drawdown that began on 25 Aug 2026 had not recovered when the test ended. - None of this is a forecast.
Your capital worksheet (copy this)
- Widest stop the EA uses: ____ pips. Max open positions: ____
- Floor equity = 0.01 × stop × pip value ÷ risk% × positions = $____
- Win rate ____ → likely worst streak over 500 trades ____ → streak DD ____%
- Bad-case DD (95th pct or ≥1.5× backtest max DD) ____%
- Dollar loss I'll accept $____ → ÷ step 4 = $____
- Margin at full exposure on my leverage $____ → margin level at that point ____% (broker stop-out: ____%)
- Real minimum deposit = the largest of steps 2, 5 and 6 → $____
Keeping it honest once the EA is live
Planning gets you to a deposit. After that, what matters is whether drawdown and exposure stay inside the limits you set.
- VRD Insight (free): a per-EA performance tracker. It groups your history by EA and magic number and shows net profit, max drawdown (from closed trades), profit factor, equity/drawdown curves and open exposure on your chart. It's read-only.
- VRD Quant ($69): a backtest analyser and EA capital calculator. It reads an MT5 backtest (run at 0.01 lots, or at your real lot for recovery and multi-leg EAs, which don't scale linearly) and suggests a fixed lot, operating capital, a reserve and an abandonment level, with Monte Carlo drawdown ranges. In other words, it does the maths in this article for your own EA. Its results are estimates from historical tests.
- VRD Guardian ($79): an equity protector and drawdown limiter. You set loss limits plus news and risky-day windows, and it steps in under those rules.
- VRD Complete ($149): gives each EA its own Passport with sized capital and limits, adds a central reserve, and includes all the modules.
Guardian makes a best-effort attempt to limit losses and protect an account in difficulty using predefined rules. It cannot guarantee that losses will be avoided or that an account can be rescued. Gaps, slippage, broker or connectivity failures can prevent rules executing. Trading carries a high risk of loss.
Start free: VRD Insight
Begin with VRD Insight, the free per-EA tracker for MT5, sold through the MQL5 Market. It splits your results by magic number so you can see which EA is earning and which is costing you (drawdown is calculated from closed trades). When you need more: 1. VRD Guardian ($79 or $25/month): loss limits per account and per EA. 2. VRD Quant ($69 or $20/month): turns a backtest into a capital plan. 3. VRD Complete ($149 or $35/month): all three, plus EA Passports and a central reserve.
Related reading
- How we score EA risk: the VRD review framework
- Running multiple EAs on one MT5 account
- Grid and martingale EA risk
Risk disclaimer: Trading forex, CFDs and other leveraged products carries a high risk of losing money rapidly and may not be suitable for everyone. Figures are illustrative and computed from stated simplified assumptions (no spread, slippage, swap or correlation), except the recovery EA worked example. That example comes from our own MT5 backtests, which are historical and hypothetical, not live results. Nothing here is a forecast or performance claim. Past performance does not guarantee future results. This is not financial advice. Virtual Risk Desk tools are monitoring and analysis software. They do not guarantee profits or prevent losses.
Frequently asked questions
What is the minimum deposit for an EA?
The technical minimum is whatever lets the EA open a 0.01-lot trade, often $100 or less. The practical minimum is the largest of the lot-floor, streak, tail-drawdown and margin checks. For a 1%-risk EA with 30–50 pip stops, that usually runs to several thousand dollars.
Can I run an EA with $100?
You can, but on most EAs the 0.01-lot floor means your real risk per trade is far above the setting. Treat a $100 account as a test of the mechanics, not a fair test of the strategy.
How much should I risk per trade with an EA?
Most risk-aware traders keep it around 0.5–1% per trade. In our illustrative simulation, going from 1% to 2% raised the chance of a 50% drawdown from about 0.01% to 7.6%.
Is the backtest drawdown enough to size my account?
No. A backtest is one historical path. Treat its maximum drawdown as a minimum, and size from a worse figure: 1.5× the backtest, or a Monte Carlo 95th percentile.
Can I scale a 0.01-lot backtest up to my lot size?
For simple single-position EAs, roughly. For recovery or multi-leg EAs, no. In our backtests of a popular MQL5 recovery EA, 0.85 lots had a drawdown about 2.5x bigger than scaling the 0.01 run predicted. Backtest at the lot you will trade.
Does higher leverage mean I need less capital?
It means you need less *margin*, not less capital. High leverage lets bigger positions open on a small account, which raises the risk of a wipeout.
How do I size a grid or martingale EA?
Size it on the full basket. Total open lots × pip value tells you how many pips against you wipe out the account. If the EA has no hard basket stop, the potential loss is unlimited.
One EA blowup costs more than the whole suite.
VRD Complete contains single-EA blowups and frees capital you have parked as idle buffers. $149 at launch pricing.