How to Hire an MQL Developer: 15 Questions That Expose Copy-Paste Coders
Most failed EA projects do not fail because of bad code. They fail because of a vague specification, a missing source-code clause, and a developer chosen on the lowest bid. The fifteen questions below — and the hiring process around them — separate a real MQL developer from a template reseller. They are the standard we apply whenever clients hire MQL4/MQL5 developers here.
Table of Contents
Why Most EA Projects Fail
Four causes sit behind nearly every failed EA project we have been asked to rescue:
- The vague spec. "Buy when price crosses the level" — without saying which timeframe, which candle close, broker time or server time, or what happens when the level gets tested twice.
- No source-code clause. The developer delivers a compiled .ex4 and keeps the code. Every future change goes through them.
- No test plan. Delivery is judged by a screenshot of a beautiful backtest — which anyone can produce by optimizing into the past.
- The cheapest bid. The lowest price is usually a template with your parameters pasted in. It compiles — that is the whole feature list.
All four are fixable with the process below — none of them requires technical knowledge, just structure.
The 15-Question Interview
Ask these before discussing price. Each one separates a senior developer from a template assembler — full engineering context is in the MQL4/MQL5 developer guide.
| Question | Why It Matters |
|---|---|
| Do you deliver full source code, or only compiled files? | Without .mq4/.mq5 files you are locked to one developer forever. |
| Show me one of your EAs — and its worst backtest, not its best one. | A developer who only shows winning curves is hiding the overfit. |
| How do you structure entry logic: inside OnTick directly, or in a separate signal module? | Spaghetti in the tick handler means every future change is a rewrite. |
| How do you handle a requote or busy error (138/146) in MQL4? | Retry logic is what keeps a live EA trading when the market is fast. |
| Explain OnTradeTransaction and when you would use it instead of polling positions. | Event-driven state sync is the core of copiers and prop-firm bots. |
| Do you support netting accounts, hedging accounts, or both? | A developer who cannot handle both will break your EA when you change brokers. |
| How do you test the EA on a 5-digit broker versus a 4-digit broker? | Hardcoded pip values silently change risk on different brokers. |
| What is walk-forward analysis and why does it matter? | It separates stable parameters from curve-fit illusion. |
| How do you prevent the EA from trading during high-impact news? | Prop firms terminate accounts for news-window trades; the filter must live in code. |
| How would you add a daily loss limit that survives a platform restart? | Persistence via globals or files is what keeps a hard stop hard. |
| What happens if the broker rejects a stop-loss modification mid-trade? | Sloppy handling leaves positions running naked at the worst moment. |
| Can you produce an MT4 and an MT5 version from the same logic? | Dual-platform delivery proves real understanding, not template copying. |
| What does your revision policy cover after delivery? | Without a defined window, small fixes become endless paid change requests. |
| Can you show a backtest matching a live statement of the same EA? | A backtest that cannot track live behavior is a fantasy document. |
| Why should a prop-firm EA cap drawdown internally instead of relying on the firm's dashboard? | Firm dashboards close accounts after the breach; internal caps prevent it. |
Questions 3, 5 and 8 are the core of modern MQL engineering — if a candidate cannot answer them, walk away.
The Specification Template
A two-page written spec beats a two-hour call — it forces every ambiguity to the surface before money moves. The eight sections every EA spec must contain:
- Strategy description. One paragraph in plain words: what the EA does and why it should have an edge.
- Entry conditions. Every rule in "if X and Y then Z" form, with parameter values and candle-close versus live-tick semantics.
- Exit conditions. Stop loss, take profit, trailing, break-even, time-based exits, signal reversals — and which one wins when they conflict.
- Risk rules. Risk percent per trade, max concurrent positions, daily loss limit, drawdown cap, and what the EA does when it hits each.
- Symbols and timeframes. Exact pairs, sessions in broker server time, and whether the EA is single-symbol or multi-symbol.
- Broker and accounting mode. Netting or hedging, 4-digit or 5-digit pricing, FIFO constraints, minimum stop distance.
- No-trade conditions. News windows, spread limits, volatility filters, and any hours the EA must stay flat.
- Acceptance criteria. The tester settings and results that mean "delivered" — see the acceptance section below.
If a developer quotes without a spec like this, their price is a guess; a serious developer will help you fill in the blanks before quoting, which is itself a screening signal. The same spec carries directly into our MT5 EA development intake process.
Fixed vs Hourly vs Milestone
Each pricing model shifts risk between you and the developer. Pick by project shape.
| Model | Best For | Risk |
|---|---|---|
| Fixed price | Single deliverables with a locked spec — indicators, simple EAs | Every change request becomes a new invoice; corners get cut near the budget end |
| Hourly | Exploratory builds where the logic evolves as you see it run | Unbounded cost; slow developers bill more for the same output |
| Milestone | Complex EAs, copiers, and multi-stage systems with review points | More administration; needs a firm spec per stage and discipline on both sides |
For anything complex, milestone is the professional default: you pay against demonstrable progress — spec accepted, build compiling, tester validation passed, delivery accepted — and neither side carries the whole risk. See our pricing page.
Source Code Rights, NDA and Ownership
The contract is where cheap projects die. Before any payment it must say: the developer delivers complete commented source (.mq4/.mq5 plus any .mqh modules); ownership transfers to you on final payment; third-party libraries are disclosed; a revision window is defined in days; and a mutual NDA covers the strategy.
Sign the NDA before sending strategy details — reputable developers expect it. A developer who refuses source delivery, or wants to "rent" you the compiled file, is not selling you software; it is a subscription to their presence. Clients rate this exact point on our reviews page.
Acceptance Testing
Acceptance criteria are agreed before the build, not argued after delivery. Four checks cover almost every EA project:
- Tester runs with stated settings. Symbol, timeframe, period, modeling quality, and seed agreed in the spec — results reported against those settings.
- Demo forward period. Two to four weeks on demo with the real broker, matching live behavior against the backtest.
- Parameter sensitivity. Nudge the key inputs a step up and down; if performance collapses, the tuning was curve-fit.
- Revision window. A defined period — typically 7 to 14 days — in which bugs against the written spec are fixed without new invoices.
An EA is accepted against the written criteria, never against a screenshot — and our team audits third-party EAs exactly this way.
Realistic Prices 2026
Market ranges as of August 2026 — below these numbers, expect a template, no source, or both.
| Deliverable | Typical Range | What It Should Include |
|---|---|---|
| Custom indicator | $50 – $150 | Source code, buffer configuration, no-repaint logic where specified |
| Simple EA | $150 – $400 | One strategy, fixed SL/TP, basic money management, source included |
| Complex EA | $400 – $1,500+ | Multi-condition entries, ATR stops, trailing, sessions, news filter, spec document |
| Infrastructure (copiers, dashboards) | $500 – $3,000+ | Deal-level sync, panels, external data feeds, documentation |
Prices below the range usually mean one of three things: no source code, no revision policy, or a template with your parameters pasted in — any of the three makes the cheap EA the most expensive one you will ever buy.
Red Flags That Cost People Thousands
Five behaviors that end badly in every market cycle:
- Guaranteed profits. Nobody can guarantee trading returns; a developer promising 95% win rates or fixed monthly gains is selling fantasy. Professional output is testable, not promised.
- No questions asked. A developer who quotes your vague idea without asking a single clarifying question is planning to build their interpretation, not your strategy.
- Refuses source code. As covered above — it converts a one-time build into a lifetime dependency.
- No revisions. "No changes after delivery" means the spec bugs are yours to pay for, twice.
- Disappears after payment. The freelance-platform special: great during the bid, silent after the invoice. Milestones and a contract are the antidote.
Frequently Asked Questions
How do I find a good MQL5 developer?
Screen with technical questions before price: source code delivery, retry handling, OnTradeTransaction, netting versus hedging, and walk-forward analysis. Ask for a worst backtest alongside the best one, and verify live matches backtest. Then agree a written specification and an acceptance test before payment.
How much should I pay an MQL developer?
2026 ranges: custom indicators $50-150, simple EAs $150-400, complex EAs $400-1,500+, infrastructure like copiers or dashboards $500-3,000+. Below the range usually means a template with your parameters pasted in, no source, or no revision policy.
Should I sign an NDA before sharing my strategy?
Yes — a mutual NDA costs nothing and takes a day. Sign it before sending any strategy detail; reputable developers expect one. Avoid developers who refuse, and never post full entry rules publicly.
How do I test the EA after delivery?
Define acceptance before the build: tester runs with agreed settings, a two to four week demo forward period, a parameter sensitivity check, and a defined revision window. An EA is accepted against written criteria, never a screenshot.
What is the difference between hiring a freelancer and an agency?
A freelancer is one point of contact at the lowest price, with availability risk if they disappear. An agency gives you a team, delivery processes, contracts and continuity at a higher price. For a one-off indicator a freelancer is fine; for systems that must keep running, an agency is safer.
Hire Once, With Structure, or Hire Forever
The fifteen questions, the eight-section spec, and the acceptance plan cost you an afternoon and save you the year a failed EA project takes. Use them on any developer — including us. If we pass, the build starts with a written specification and ends with your source code.