How Much Buffer Does Your Account Need Before Running a High-Win-Rate EA?

The curve you want (the smooth one, climbing month after month, 90%+ win rate) comes with a bill attached, and the bill has a specific size. Before anyone runs a high win rate EA, mine included, they should know that number: the account buffer that survives the strategy's worst stretch with room to spare. I give it to my own buyers before they pay, I have talked people out of purchases with it, and this post is the full version of that conversation.

Because here is the pattern that fills refund queues and review sections across every marketplace: a genuinely solid high-win-rate EA, bought by someone whose account could not afford its rare bad week, switched off (or margin-called) at the exact bottom. The EA gets the one-star review. The real culprit was a missing paragraph on the sales page: this one.

Why High Win Rates Mean Deep, Rare Drawdowns

This is mechanics, not scandal. A strategy that wins 90%+ of its trades is, by construction, taking many small wins and accepting that its losses (rare by design) will be larger when they land. Often such systems also manage losing positions actively (holding, scaling, working the position back) rather than taking every small stop, which is precisely what produces both the high win rate you like and the occasional deep equity dip you must be sized for. There is nothing hidden about this trade-off; it is the same asymmetry read from two directions. What gets hidden is only ever the second half.

So the honest translation of "92% win rate" is: months of steady green, punctuated by rare stretches that will test your account and your nerves, and the strategy's profitability depends on surviving both. Whether that trade-off is right for you is exactly what the buffer math decides. (And if a vendor's smooth curve comes with no visible dip at all, you are looking at a different problem: what "low drawdown" numbers hide, or a position-management style that parks the risk in the floating column. Both are readable on a verified track, once you know where to look.)

The Buffer Formula (Pen, Paper, Three Minutes)

The general framework for any robot is in the one number that decides robot profitability. For high-win-rate EAs specifically, run it with these steps:

1. Take the verified maximum drawdown from the LIVE track. Not the backtest. Include floating drawdown if the tracker shows it: for position-managing strategies, the floating number is the real one.
2. Multiply by 1.5. The worst stretch on record is the worst so far. Sizing to survive only history is planning for the past.
3. Convert to money at your risk settings. A -17% historical worst becomes a planning number around -25%. On a 5,000 account, that is 1,250 underwater at the deepest point. On 2,000, it is 500 that must not make you flinch.
4. Apply the two survival tests. Financial: does the account stay far from margin trouble at that depth, at your lot sizes? Emotional: can you watch that number, possibly for weeks, without switching the EA off? The second test fails more accounts than the first.

If both tests pass, you have consented to the real cost of the curve, and the strategy's math can work for you. If either fails: smaller size, or no purchase. Those are the only two honest outcomes, and the second one is fine.

A Worked Example, From My Own Shelf

Let me run the math publicly against the product this conversation usually involves. My GBPUSD system's verified Myfxbook history: a 92% win rate across 1,179 trades (1,092 winners), +165.72% over 13 months, profit factor 2.18, and there it is, the number this whole post exists for: a maximum drawdown of -16.79%.

Apply the formula. Planning drawdown: -16.79% × 1.5 ≈ -25%. On a 5,000 account at standard settings, that is a worst-case stretch of roughly 1,250 underwater; on a 3,000 account, about 750. The buyer I want is the one who reads that, nods, and sizes accordingly. The buyer I turn away (and I do, in writing, before they pay) is the one with 800 of margin who needs this month's profits to matter. Not because the EA would not win for them (the math is the same math) but because their account cannot afford the losing stretch that the math requires them to sit through, and a forced stop at the bottom turns a working strategy into a personal loss.

The same curve everyone sells. Plus the number they don't.

DoIt GBP Master is that GBPUSD system: 92% win rate, 13 months of verified public history, and the -16.79% worst stretch printed where you can see it, because you cannot consent to a cost you are not shown. If your buffer math says yes, it is the premium EA on my shelf. If it says no, do not buy it, and I mean that.

The Pre-Purchase Checklist (Apply It to Everyone, Including Me)

• Live verified track, six-plus months, losses visible. No live track: no purchase, whatever the win rate claims.
• Maximum drawdown located, floating included, and multiplied by 1.5 into a planning number.
• Planning number converted to money on your actual account at the vendor's recommended settings.
• Both survival tests passed, and the emotional one answered honestly: the strategy only pays the owners still running it at the bottom.
• Vendor asked directly: "What buffer do you recommend for the worst stretch?" A concrete answer with a number is a green flag rare enough to be a buying signal in itself. Evasion is your answer too.

The Honest Close

High-win-rate EAs are neither a scam nor a free lunch; they are a specific trade: steady small wins in exchange for rare stretches that demand capital and nerve. The vendors who show you only the first half are not selling you a strategy, they are selling you the first act of one. The whole strategy includes the bill, and the buffer is how you pre-pay it on your own terms instead of the market's.

Run the three-minute math before any purchase. If a seller has done it for you, unprompted, with their own worst number in plain view, you have found the rarest thing in this industry: someone selling to adults.

The weekly view of what my systems are actually printing (green weeks, red stretches, and the buffer math updated live) is in the DoItTrading newsletter. No airbrush, by policy.

Frequently Asked Questions

How much money do I need to run a high win rate EA?

Derive it, never guess it: take the EA's verified maximum drawdown from its live track, multiply by 1.5 as a planning assumption, and convert that percentage into money at the recommended risk settings. Your account needs to absorb that stretch while staying comfortably clear of margin trouble, and you need to be able to watch it without switching the EA off. As a worked example, a system with a -16.79% verified worst stretch implies planning for about -25%, which on a 5,000 account means tolerating roughly 1,250 underwater at the deepest point.

Why do high win-rate EAs have deep drawdowns?

Because the two properties are the same asymmetry read from opposite ends: many small wins are purchased by accepting rarer, larger losing stretches, often via active management of losing positions rather than constant small stops. That design produces both the smooth months buyers love and the occasional deep dip they must be sized for. It is not inherently dishonest; hiding the second half of the trade-off is.

Is a 90% win rate EA safer than a 50% one?

No, just differently shaped. The 90% system feels safer for months at a time and concentrates its risk into rare, deep stretches; the 50% system distributes small losses constantly and feels worse week to week while often carrying shallower worst cases. Safety lives in one place only: whether your account is sized to survive the specific system's verified worst stretch with margin to spare. Judged by that standard, either shape can be safe and either can be lethal.

What buffer should I keep beyond the EA's drawdown?

Plan for 1.5× the verified historical maximum drawdown as your baseline, more if the track is short or the strategy manages losing positions (where floating drawdown can exceed the closed number). The record's worst stretch is only the worst so far, and drawdown records tend to break at the least convenient moment. Whatever number you land on, write it down as money, not percentage, and confirm you would genuinely leave the EA running at that depth.

Should I stop my EA during a losing streak?

If you did the buffer math before funding, the losing streak is inside the plan and the answer is no: switching off a positive-expectancy system at the bottom converts a temporary, budgeted drawdown into a permanent personal loss, and it is the single most common way buyers lose money on working EAs. If you find yourself wanting to stop it, the real problem is that the sizing was never survivable, and the fix for next time is a smaller size or a bigger buffer, decided before the money, not during the pain.