Every strategy carries its exact rules, an annotated chart of the setup, the research behind why it should work at all — and the conditions under which you should not take it.
Showing 71–80 of 127 strategies
A Donchian channel is the highest high and the lowest low of the last N bars. There is no averaging, no deviation, no smoothing — it is the most mechanical measure in this book, and the hardest to argue with about what it says.
An envelope is a moving average with a fixed percentage band around it — not a standard deviation, not an ATR multiple. A half-percent envelope is half a percent whatever volatility is doing.
A pair has an average daily range, and most days it does not travel much further than that. When price has already covered its whole average range by mid-session, the rest of the day is more likely to be spent going sideways or giving some back than extending.
In an efficient market, the act of trading leaves a trace. Buyers lift the ask and sellers hit the bid, so observed prices alternate between the two sides even when nothing about the instrument has changed. Roll showed this produces negative serial correlation in consecutive price changes — and that the size of it measures the spread itself.
Over very short intervals, price does not move because of a pattern. It moves because the flow of orders at the best bid and the best ask is lopsided. Cont, Kukanov and Stoikov measured that directly and found a linear relationship between order flow imbalance and the price change that follows, with the slope set by how deep the book is.
Every scalping strategy has the same denominator: the spread you pay. That spread is not constant through the day — it is narrowest when the most participants are in the market, and the range available to trade is widest at the same time.
A scalp targeting six pips on a pair with a one-pip spread and a round-turn commission worth another 0.7 pips is not a six-pip trade. It is a 4.3-pip trade, and the stop has to absorb the same 1.7 pips on the other side.
Two accounts running an identical bot on the identical strategy will produce different results, and on a scalping timeframe the difference can be larger than the edge. The cause is execution: how the order is routed, how long it takes, and what happens when the market moves while it is in flight.
Ranges do not narrow at random. A market that has stopped moving is a market where the two sides have run out of disagreement, and that state does not last — the next expansion arrives, usually soon, because volatility clusters.
Two ways of measuring the same width disagree in a useful way. Bollinger bands are built from standard deviation, which collapses fast when price goes quiet. The Keltner channel is built from average true range, which collapses slowly.
The mechanism has peer-reviewed, FX-specific research behind it.
The mechanism is sound, but at least one step of it is untested.
Practitioner literature only — no published research supports the mechanism.
It rates the mechanism — whether there is published research explaining why the market should behave this way. It does not rate profit. Strong does not mean the trade will work. None of these have been run on our own account, and no page here shows a backtest figure.
One instrument, one timeframe, every rule visible on the chart.
Several conditions have to line up, and execution starts to matter.
Needs an outside data feed, a basket, real order flow, or an unusual tolerance for lumpy losses.
The level describes what it takes to run the strategy, not how well it works. Several Beginner pages are better evidenced than several Professional ones — a basket carry trade is Professional because it needs eight currencies and a swap table, not because it is cleverer than buying a pullback.
Reading the library is free. A plan is what lets you turn one into a robot.
Trading carries risk and most retail accounts lose money. Nothing in this library is financial advice or a recommendation to trade. The charts are illustrations of the rules on constructed candles — they are not records of trades and not backtests.
