Odin
Tuurt Trading's third robot, for MetaTrader 5, for sale under licence. It is the MQL5 port of Tuurt's Python gold bot, built so the Strategy Tester can measure it on real ticks. We measured it twice and both came out badly: all four variants of the 1.x engine lose, and the 2.1.4 engine breached a funded-account challenge's total drawdown limit on trading day 13. Both tables are below in full, unedited.
- Platform
- MetaTrader 5MQL5, Windows terminal
- Status
- For saleUnder licence, one key per account
- Trades measured
- 1,692On the base arm, no management
- Challenge measured
- Broken on day 132.1.4 engine, FTMO phase-1 rules
An MQL5 port of Tuurt's Python gold bot.
Odin is not a new robot built from scratch: it's the MQL5 translation of the signal core Tuurt already runs in Python on gold. The design doesn't change in the port — what changes is the engine that measures it.
The architecture, as designed, is top-down: H4 defines the trend, H1 confirms the level, M15 gives the timing. Three timeframes, each with a distinct job — the largest sets the bias, the middle one validates the entry level, the smallest fine-tunes the exact moment to pull the trigger.
The Python bot sees 48 moments a day. Its backtester had never been checked.
The Python bot samples every 30 minutes — it sees about 48 moments a day, not every tick. And its own backtester had never been checked against real data.
Porting the core to MQL5 lets the MetaTrader 5 Strategy Tester model it on real ticks, and so answer the only question that matters before any other: does the signal have an edge at all.
Four arms, same signal. None wins.
Strategy Tester run from August 13, 2026, on real ticks (confirm_ticks.csv).
core is the signal alone, with no management of the open position. The other
three are variants of one rule: once profit reaches a given percentage of the target, if price
comes back to that level the position closes and collects that minimum — they differ in the
threshold and in how the level arms.
| Arm | Profit factor | Net | Max drawdown |
|---|---|---|---|
| core Signal alone, no position management | 0.99 | −3,026 | 27.8% |
| piso30 | 0.92 | −24,140 | 52.4% |
| clamp20 | 0.93 | −22,971 | 52.7% |
| arm20 | 0.92 | −24,242 | 55.8% |
Four variants, none wins. A profit factor below 1 means that, over the course of the run, more money went out than came in.
Management does not rescue a signal with no edge
The three managed arms — piso30, clamp20, arm20 — don't just lose: they lose more than the bare signal, and roughly double the maximum drawdown. Putting a guaranteed minimum on a winning position doesn't make up for the fact that the core already loses over 1,692 trades.
The core arm ran 1,692 trades; the three managed arms, around 4,800 each.
The 2.1.4 engine against FTMO phase-1 rules: broken on day 13.
The table above measures the 1.x engine. This one measures 2.1.4, and not against a profit target of ours but against the rules of a funded-account challenge. Strategy Tester run from 1 September 2026, on 100% real ticks, window 1 January to 1 August 2026. Simulated deposit of 100,000, leverage 1:100.
| What was measured | Result | Firm's limit |
|---|---|---|
| Verdict Limit breached on 15 January 2026 | Challenge failed | — |
| Days traded before breaching | 13 | — |
| Worst total drawdown | 10.17% | 10% |
| Worst day | 2.99% | 5% |
| Final equity On a 100,000 simulated deposit | 89,826.27 | — |
It died by accumulation, not by one bad day: it never came close to the 5% daily limit, and still ate the full 10% in thirteen trading days. The result is identical under both plausible day anchors (UTC+1 in winter, UTC+2 in summer).
What this run does not say
The rule values are the ones we assumed, not the ones we verified. Profit target 10%, max daily loss 5%, max total loss 10% of the initial capital and static, day anchored at 00:00 Prague time. They have not yet been checked against a real FTMO dashboard.
It is a single start date. The challenge began on 1 January 2026 and was over by the 15th, so the remaining ≈ 6.5 months of collected data never bore on this result. A real challenge starts the day you buy one, so "does it pass" is a question about many possible start dates, and this measures exactly one of them.
It is Strategy Tester, not real money. None of these figures comes from an account holding client funds.
Odin ships with these tables in front of you, not in spite of them.
Odin is for sale under licence. What you have just read is not a draft, nor an earlier stage we are about to correct in the next version: it is the most adverse measurement we hold of this robot, and it stays on its product page because it is exactly what you would be buying.
This section used to be titled "why it's not for sale", and it drew from these same numbers the conclusion not to ship it. The decision changed; the numbers did not. Both tables are still here, with not one figure edited, because a robot that is sold needs that information more, not less: someone paying for a licence has more claim on it than someone who was only looking.
What you buy is the binary, the key tied to your account number, and the configuration reviewed with you. You are not buying a result. Of the four 1.x variants measured on real ticks the best one loses less — none wins — and the 2.1.4 engine, against FTMO phase-1 rules, breached the total drawdown limit on day 13. If that makes you close the page, the page did its job.
Buying Odin is not buying the funded-account service
They are two different things with different terms. The licence on this page hands you the robot for your own broker account: you run it, you configure it, and the risk on that account is yours from beginning to end.
The funded-account service is a separate product: there Odin runs on your challenge account with the firm's daily loss limit configured inside it, you pay half on signing and half when the account becomes funded, and it has its own page with its own fine print — including that as of 1 September 2026 no challenge has been passed with it. Buying this licence does not include it, and signing up for it is not buying this licence.
Risk disclosure
Odin is sold under licence, and no measurement published here is profitable. The numbers above describe Strategy Tester runs, not a live account, and none of them is a forecast of what it would do on yours.
Buying this licence is not signing up for the funded-account service. They are two different products with different terms; the funded service has its own page and its own fine print.
The instrument measured is high-risk. The bot trades leveraged gold via CFDs; it can make you lose the entire capital of an account.
The measured result is negative on all four variants. A profit factor below 1 means a net loss over the course of the run; no configuration on this page is profitable today.
And against the rules of a funded-account challenge, the 2.1.4 engine broke it. It reached 10.17% total drawdown — past the 10% limit — on trading day 13, with a worst day of 2.99%. Those rule values are the ones we assumed for FTMO phase 1 and are not verified against a real dashboard.
No number on this page comes from real money. All of it comes from the Strategy Tester run on historical ticks, not from an account with client funds.
Nothing published here is investment advice nor a promise of future profitability. No past result, measured or projected, guarantees a future one.
Trading leveraged CFDs carries a high risk of losing money rapidly, including the entire deposit.