Drawdown — Surviving the Losing Streak
Streak maths for a working edge, the recovery curve's asymmetry, the 1%-versus-5% proof across 20,000 runs, and the ladder written before the streak arrives.
Hands-on modules with live charts, tap quizzes and playbooks — a guided route from your first candle to a complete strategy.
Free, interactive modules for learning to trade — candlestick reading, support and resistance, VWAP, session timing and momentum — with a guided study path from your first candle to a full strategy.
Streak maths for a working edge, the recovery curve's asymmetry, the 1%-versus-5% proof across 20,000 runs, and the ladder written before the streak arrives.
The wipeout table from 1:5 to 1:30, a bar-by-bar walk into a margin call with every number computed, volatility drag, and the sizing that makes leverage boring.
A trader with no edge passes a typical evaluation 28% of the time by luck; a real +0.15R edge passes 53% and is then closed out within 1.7 months on average, because a 5% drawdown at 1% risk is a five-loss leash. For a measured edge the fee buys a real asymmetry; for anything less, a certificate.
The journal said +£1,450; the broker statement said −£144; the £1,594 between them is spread, slippage and financing nobody logged. Every statement, every deposit and withdrawal with its bank record, the broker's trade export, and one reconciling sum a month. Not tax advice — the records that make it possible.
A thirty-day at-the-money call on gold is worth about 41 points across forty thousand computed outcomes. The buyer profits 34% of the time; a two-sigma move pays 5.2× the premium; with no move the option loses 29% of its value by half time and 59% with five days left. A bet with the stop built in — and rent on the time.
Three computed years of a working strategy from £2,000: reinvest everything and the account reaches £4,714; withdraw every month's profit and it shrinks to £944, because losing months are never refunded. “1% a day” needs +1.0R of expectancy; a real edge earns +0.09R.
A working strategy has a losing month 36% of the time and two in a row 13%; a month's reading runs from −0.26R to +0.44R around a true +0.09R. The month judges the process; the rolling hundred judges the strategy. The meeting decides stop, step or continue — and never mixes them up.
Two weeks on the chart, four on a demo with a written plan, seven live at quarter size: ninety journalled trades of one strategy. They measure an edge to ±0.15R — a real edge reads negative 27% of the time, a broken one reads positive 0.7%. Enough to catch a broken strategy, not enough to crown a good one.
A long and a short in the same market is being flat at 32× the cost of closing. Hedging gold with silver at 0.8 correlation removes 64% of the risk and costs ~4R a month; simply halving the position removes more, for nothing. A smaller position is the hedge that's free.
Four hundred computed release days: the fifteen minutes after the number held 32% of the day's range at a spread five times normal. A 2-point stop was taken 53% of the time, an 8-point stop 2%. The 15-minute direction held to the hour 48% of the time — a coin. Weather, not a setup.
£1.56 a trade on 0.05 lots of gold: £9,766 a year for a scalper (nearly five times a £2,000 account), £1,172 for a day trader, £244 for a swing trader. On a 2-point stop the bill is 0.20R of every trade; on a 25-point stop, 0.016R.
Ten computed years on a market drifting 7% a year: buy and hold +76% for ten hours a year; the same index on a CFD −8%; trading with no edge −51%; with a +0.10R edge +36% for five hundred hours; with +0.20R +264%. Two accounts, never one.
After 30 trades, 9% of no-edge traders look like +0.30R; after 100, 1%. The lucky ones who doubled at thirty ended with a 52% drawdown. Waiting costs a real edge about two points. Four conditions, a date, and a half-per-cent step.
With a real +0.15R edge, eight losses in a row happened in 56% of hundred-trade samples and ten in a row in 26%. An eight-run is only 1.9× as likely from a coin — a streak is evidence, not a verdict. Audit the trades, keep the size, decide at a hundred.
Twenty thousand computed weekends: a half-ATR stop was gapped through 19% of the time, a 2-ATR stop 1%. The worst gap cost £735 on a 0.05-lot gold position. A guaranteed stop costs several times the expected loss — and caps the one that would end the account.
The spread is 5% of a typical hour's range at 14:00 and 38% at 22:00 — the same trade at seven times the price. The first minutes of an open, the rollover window, holidays and Friday afternoons are worse than any hour. The trading day is shorter than the market's.
A UK nine-to-five can watch 34% of the day's movement, 20% of it in the evening; 42% happens at work. The evening produced about five hourly setups a week — one a night. Three shapes that fit a job, and the phone under the desk that doesn't.
A 2R plan earning +0.09R a trade. Stops widened on three losers in ten: +0.04R. Half the winners taken at 1R: −0.10R. Both: −0.16R — a £1,800 swing a year at £20 risk. The edge lives in the second R of the winners.
A plan that returns +36% a year. Revenge trades after two losses in five: −21%. Double the trade count with no extra edge: −24%. Both: −56% with a 65% drawdown. 62% of all trades follow a loss — the moment after a loss is the ordinary moment.
Doubling from £20: a £2,000 account funds six losses; the seventh needs £1,280. 85% of computed accounts ruined within a year, and the survivors ended near £4,960 and posted the screenshots. Two adds to a losing gold position cut the room to a forced close from 392 points to 41.
A 0.3-point spread is 104% of a typical 1-minute gold bar, 37% of a 5-minute one, 7% of an hourly. At 2R a 1-minute scalper needs 68% winners just to break even; a daily trader, 34%. Start on the hourly.
Balance is what you had, equity is what you have, free margin is how long you've got. On a £2,000 account gold can move 392 points against 0.05 lots before a forced close — and 8 points against 0.20 lots.
Two thousand computed first years of a steadily improving beginner: at 1% risk the typical account ends down 17% with a 36% hole on the way; at 3% risk half lose three quarters. A good year one is a full journal and an intact account.
Dealers, funds, hedgers and other retail traders — doing their own jobs, not fighting you. A broker collects £1,350 in spread from a careful client over three years and £938 from a reckless one gone in a month.
One point of gold on a standard lot is about £78. Five points against you is £391 — 20% of a £2,000 account. Sized to 1% risk, that same stop is 0.05 lots. The lot decides everything.
The same week turned 710 times a day at one minute and 12 at one hour — same prices, different amount hidden. A linear axis draws the same 10% move ten times taller at 1,000 than at 100.
Five related trades at 1% are 1.2 effective bets. All five lost together one day in three, against one in twenty when unrelated — a 1-in-20 day of −5% vs −2% for a single 2% trade. Cap heat per theme.
Fixed lots risked 2.3% on the worst day. The 1% rule with a stop inside the noise lost 40%+ in 94% of runs. A half-ATR stop was hit 70% of the time, a 2-ATR stop 40% — stop distance is a strategy.
The fade: +0.78R in ranges, nothing in trends. The breakout rule: +0.14R in trends, −0.65R in ranges. A twenty-bar read gets the regime right 73% of the time — read first, then pick the rule.
From any anchor — swing, breakout, news, or a random bar — price came back into value ~85% of the time and to the busiest price six times in ten. Anchor to the event you're trading; the behaviour is the same.
Five anchors against a random bar: all revisited ~86%, all held about half the time. Even a line defended 50% of the time held 55% vs 55%. Choose the anchor by the question, once.
Two-ATR bricks cut direction changes from 499 to 32 per thousand bars. Every reversal is two bricks late, and the same rule earned −0.05R, +0.03R or +0.13R depending only on the brick size.
Every line is a midpoint of recent prices. It pointed the same way as a 9/26 EMA pair 90% of the time; the cloud held as support 71% against 54% for a random band. One fact, drawn five ways.
Buy the 20-day high, exit the 10-day low: +0.24R a trade from 25% winners averaging +4R, a 17-loss run, and a 46R drawdown — £912 on £2,000. It works; most people can't hold it.
A 500-name scan hands you 15 chance matches a day beside 10 real ones, and the ratio is 40% at any width. Five names, a fifteen-minute prep, and a list that's allowed to be empty.
A 3:1 order book put price on its side 56% of the time five ticks later, 52% at a hundred — real, small, fast-fading, and hundreds of hand-placed trades to prove.
One default lot on a £2,000 account is ~£100,000 of exposure — a 2% move erases it. The ten-minute setup, and market vs limit orders with a break-even.
The textbook promises 95% inside the 2σ band. Measured: price was outside 15% of the day even with tame moves — and fading every touch earned +0.01R.
The early push carried on 64% of the time. Fading blind paid +0.01R; waiting for a close back through the open paid +0.22R.
Against its neighbours the 200 was indistinguishable — a 3-point spread. As a direction filter it turned −0.10R into +0.12R in a trend, and made a range worse.
88% of trades that ended +2R were showing a loss first, and 42% of candles are red in a month that made money. Four fixes that take five minutes.
Con-tango costs, back-wardation pays back. 0.5% a roll compounds to 16.5% over three years — £10,000 becomes £8,349 with the price unchanged.
The reversion is real — 58% within two years. But 55 years produced only 11 episodes, and a career gives about 4, leaving an honest range of 15–85%.
£1.85 a night on a £10,000 position, £5.55 on Wednesdays — and an edge of £6 a trade fully consumed after 3.2 nights.
16% of no-edge traders looked skilled after 30 trades. The four tests that settle it, and the ~96 trades needed to be sure.
90% winners losing £480 over 200 trades against 35% winners making £1,600 — plus the one-line formula for the win rate you actually need.
Win rate, average win and loss, expectancy, profit factor, drawdown — and the same strategy measuring anywhere from 25% to 65% after 20 trades.
What each of the five is actually for, the reading order, and the warnings the back covers leave off — including the one about position size.
Higher high, first lower low, flat ceiling. Measured: a BOS continued 65% of the time, a CHoCH was right 50% — a coin flip.
22 different valid lines fitted one small chart. The four rules that fix that, plus the computed case for waiting: 42% to 69% win rate.
Measured: the SMA runs 9 bars behind, the EMA 7, the HMA just 2 — and that speed cost 167 false crosses and turned a profit into a loss.
A seasonal chart built entirely from coin flips — and the maths showing 73 calendar tests hand you about 3.7 convincing patterns made of nothing.
Three words instead of the jargon: the busiest price, the value band, the quiet prices — and why price travels about six times faster where nobody traded.
Yesterday's high, yesterday's low, today's open, the nearest round number — and the computed proof that fifteen levels mean the same as none.
A third of winners shake you off first. The three responses, computed — giving up loses, chasing loses more — and the clean-reclaim rule that makes one retry pay.
Five reasons, one point each, scored before entry — and the computed month where taking everything loses £102 while taking only the best dozen wins £96.
The gate that says you're ready, the quarter-size trial with a worst case of £200, why a red first month proves nothing, and the ladder up to full size.
The coin-flip trick behind perfect records, the compounding test that breaks any claim, the leaderboard illusion computed on 500 no-edge accounts, and the shark filter.
What each wrapper legally is, the computed cost of a day, a month and a year in each, the crossover at about a day, and the wrapper-to-holding-period fit.
The pass-or-fail safety floor, fee menus priced against one identical month, the size where the cheapest broker flips, and the adverts to walk away from.
The three jobs of starting capital, the same edge computed at three balances, the flat-fee tax on small accounts, and the honest wage number nobody advertises.
Alert placement that buys you measured lead time, the bracket order that carries the plan, what to automate and what never to, and the limits worth respecting.
The pre-market twenty minutes in order, a watchlist kept small on purpose, trading the windows and standing down between, and the shutdown that feeds the journal.
Good losses versus bad, the five-question autopsy, cause buckets from a 40-trade journal, and the measured cost of the trades that broke the rules.
Constraints before charts: what a fixed spread really costs on a fast chart, what changes between timeframes on one series, and why switching markets restarts your learning.
Why orders cluster at big figures, the three ways price meets one, the strength hierarchy, and why your stop should never rest just beyond a round number.
Measured proof that falls are faster and wider than rallies, the short squeeze, borrow and funding costs, and the asymmetry of a loss with no floor.
Common, breakaway and exhaustion gaps: which fill and which run, the fade and its real failure mode, and letting the first thirty minutes classify it.
Positive and inverse relationships with computed coefficients, the hidden concentration when three trades are one trade, and reading a correlate that stops agreeing.
Accumulation and distribution without the jargon: the climax and rally that build a range, the spring, the sign of strength and the last point of support.
Flags, triangles and head and shoulders taught as supply and demand — the argument each one records, the measured move, and why most pattern trading fails.
Fixed targets versus trailing, scaling out, the real cost of a breakeven stop, and letting your own trade data choose the exit rule.
Why win rate tells you almost nothing: a 70% strategy that loses money, a 40% one that prints, the break-even table and what costs really take.
Defining the box, breakout entries with a held retest, the measured move target, failed breaks that fade, and choosing your window.
The confluence zone where the 20 EMA meets the fib pocket and prior structure, the trigger candle, stops, targets, and when a pullback becomes a reversal.
How a three-candle imbalance forms, the body-close mitigation rule, trading the retest, and the inversion flip when a gap fails.
Why the first spike traps both sides, spread blowouts, the stall-and-fade play, genuine trend days, and the before/during/after playbook.
Where stop pools build, sweep versus genuine break, double raids, session extremes, and entering on the reclaim close.
Manual testing rules, why 20 trades prove nothing, curve-fitting and hindsight traps, forward testing and the ladder to going live.
Why the mind misfires, the emotional cycle, tilt and revenge trading, FOMO, and the structural fixes that beat willpower.
Relative volume, confirmed versus hollow breakouts, drying pullbacks, climax exhaustion and what tick volume really measures.
How timeframes nest, the three-screen method, aligning entries with the higher-timeframe trend, conflict traps and timeframe shopping.
The five-section plan, circuit-breaker rules, the journal in R with the Y/N column, and the weekly review loop that compounds.
The exact P/R1/S1/R2/S2 arithmetic, range-day fades and trend-day flips at the levels, confluence and the days the map fails.
How swings form, reading HH/HL sequences, break of structure, change of character, and turning the skeleton into entries and stops.
Defining the box, fading the edges on sweep-and-reclaim triggers, the mid rule, width maths and the handoff when the range breaks.
Drawing anchors correctly, the golden pocket, confluence with real levels, extension targets and when the ruler lies — fibs without the mysticism.
Market, limit and stop orders, the spread, slippage and gaps, margin mechanics and the true round-trip cost — the plumbing between the click and the fill.
Confirmed breaks, sweep-style retests, the failed-retest trap, stops behind the wick and measured targets — the gentlest complete strategy.
What moving averages really measure: SMA vs EMA, lag, dynamic support, honest crossovers and the settings that matter.
Position sizing, the 1% rule, R-multiples, drawdown maths and surviving losing streaks — the module to master before any strategy.
The trend-day pullback to VWAP: the regime filter, touch-and-reject entries, the test-count rule, and the flip when the line breaks.
Spotting the volatility coil with Bollinger BandWidth, surviving the head-fake, and trading the expansion with defined risk.
The four divergence types, how to draw them correctly, grading the signal, and turning warnings into confirmed reversal trades.
What RSI measures, how to read overbought and oversold properly, the 50-line, zone exits and divergence — momentum for beginners.
Reading the 9/20/50/200 ribbon, the entries that work at each line, and how to exit with the trend still intact.
What the Average True Range measures, and how to use it for stop placement, position sizing, entry filters and trailing exits.
What actually creates support and resistance levels, and how to enter profitable trades at them.
A simple EMA-based gold strategy you can run on eToro without leverage.
What each indicator is, how to use them well, where they differ, and how to combine momentum with price action.
How professionals trade around VWAP: reading the standard-deviation bands, band fades, and the indicators that pair with it.
The best windows to trade gold, when to stand down, and what session manipulation looks like on the chart.
How execution algos work their orders into the market, and how to find entries alongside institutional flow.
How Heikin Ashi candles are built, what they signal, and how to use them in a live trade.
The 30-minutes-a-day refresher-plus-depth training programme on reading candlestick charts.
Every module in the order it compounds, folded into ten phases you can actually finish. Pick a phase, work through it at a module a day, and come back for the next.
Every phase is a short course with its own finish line. Open one to see its modules in study order — new modules join a phase, so this page never gets longer.