LY Traders AI LabLY Bots ← Back

LY BOTS — JOURNAL

The Median Losing Streak Is Five. Can You Sit Through It?

A trader we know has been "working on it" for three years. In any given fortnight he changes timeframe, switches market, adds an indicator, removes it, doubles his size after a win and halves it after a loss. Every change feels like progress. None of it can be measured — because you cannot get a reading of anything while six variables are moving at once. The failures he attributes to discipline are, more often, an unreadable experiment.

This piece is the opposite move: fix every variable, run a single setup for twenty-one working sessions, and then — only then — look at what your own execution actually produces. Not a habit programme. A trial, with a defined end and a decision waiting at the end of it. Keep it or drop it, on evidence.

The Twenty-One Day Claim, Honestly Priced

The "21 days to build a habit" line is folklore. The best-known research on the question — Lally et al., University College London, 2010 — found new habits took between 18 and 254 days to become automatic, with a median around 66. Twenty-one sessions will not make anything automatic, and anyone selling you that number is selling the folklore.

What twenty-one sessions of a fixed setup can do is different and better: they produce a baseline. Most traders have never seen one. They have opinions about their win rate and feelings about their discipline, but no clean measurement of either, because the setup never stays still long enough to be measured. The surprise, for anyone who completes this honestly, is not the market — it is finally seeing their own numbers. We claim nothing about what those numbers will say. That is the point of measuring.

The Constraint Set

One of each. Not because simplicity is virtuous, but because a fixed frame is the only frame that produces a readable result.

No trading or coding experience required. No news-watching. No screen-staring — the daily time cost is minutes, not hours. No capital is strictly necessary: a demo account works, though be aware that paper trading rarely produces the emotions real money does, and this experiment is partly about what you do under emotion. Capital that could go to zero without touching your lifestyle is the honest sizing, if you go live.

The Tools, and the Toolmaker's Own Warning

Everything named here is free and none of it is ours. We have no affiliation or partnership with any provider named in this piece.

Charts: TradingView's free tier (desktop and mobile, same login). Execution: MT5, cTrader, or any CFD-capable platform your broker provides. The indicator class: a trend-line indicator that flips colour on direction change. One popular free example is Xtreme Trend by KachraStocks on TradingView — and its author's own disclaimer does the risk-warning better than we could: "This is no holy grail. Works well in trending market and hell in sideways." That sentence is the whole risk profile of trend-following in one line, from the person who built the tool. His second warning matters just as much: signals are only valid on candle close — act on the close, never mid-bar.

The setup is one-time: chart to 30 or 45 minutes, add the market to your watchlist, add the indicator, set its colours so long and short are unmistakable, and set an alert through the platform's mobile app so the signal finds you instead of you watching for it.

The Process

The framework is stop-and-reverse. When the indicator flips from short to long, close the short and open a long. When it flips from long to short, close the long and open a short. One position, always in the market, in the direction of the last signal. One to three signals a day on these timeframes is normal.

Risk is defined before the first trade: the stop sits roughly 1% away from entry. On a $20,000 notional position that 1% move is about $200 — 2% of a $10,000 account. Fixed lot size, every entry, regardless of how confident the last signal felt.

The Arithmetic — Both Halves

All figures below are for calculation purposes only, not a recommendation, and the instrument's live price will differ. A reference price of $64,885 on BTCUSD, a $10,000 account:

Now the half these setups always leave out: the losing streaks. In roughly 60 trades with a 50% win rate, the median worst losing streak is 5 trades; one time in ten it reaches 8; one time in a hundred it reaches 11. At 2% of equity per trade, streaks alone put the account 10–16% below its high-water mark — without the framework being broken, and without you doing anything wrong. If a 16% drawdown on the trial capital would push you into changing the rules mid-experiment, the experiment is mis-sized, not the market unkind. Size it down until the arithmetic can run its course without your interference.

Measurement Without a Journal

No daily writing. The platform keeps the record: every entry, exit, timestamp, and result is in the account history. At the end of the twenty-one sessions, export the statement — or open it on a screen with someone you trust — and read three numbers off it:

  1. Win rate — the proportion of closed trades that were profitable.
  2. Average win versus average loss — the ratio that actually decides whether the system earns.
  3. Maximum losing streak — the number that decides whether you survive long enough for the ratio to matter.

Feed the statement to an AI assistant and ask for the cost breakdown, trade timing, and long/short split if you want a second pair of eyes on it. The record is machine-generated, which is exactly why it beats memory: it cannot flatter you.

Then Decide — On Evidence

At session twenty-one, the trial ends and the decision begins. If the three numbers show a positive expectancy, you now have evidence — not motivation — that a fixed, boring process beats your old six-variable churn, and extending it is a decision, not a hope. If they show the framework bled in sideways conditions, you have learned the same lesson the indicator's author warned about, at measured cost, in twenty-one sessions instead of three years. Either outcome is a win over not knowing.

Most people will not finish. The dropout is not a failure of the experiment; it is data about how much of the "discipline problem" was really a boredom problem. And for the ones who do finish, the twenty-one days will not have formed a habit — the research says that takes months. They will have formed something rarer: a number you can trust about how you actually behave when the rules are fixed and the money is real.

The market changes. The rules don't.