Glossary

Algo level (deviation quantile)

An algo level is a deviation of price from an interval's open that a set share of candles did not exceed in history, for example 70% or 90%.

It is computed separately for each time of day and asset, up and down, for 15-minute, hourly and four-hour intervals. The 90% level is a deviation price exceeded in only 10% of cases.

More in the video The math of extremes: how algo levels are computed. See also: Quantile, Cheap and expensive, Day open.

Quantile

A quantile is the value below which a set share of the sample's observations lies.

Algo levels are quantiles of the distribution of deviations from the interval open: level 60 means that in 60% of observations price at that time was below it and in 40% above.

More in the video Algo levels: how they are computed and how to tune them. See also: Algo level, Sample.

Cheap and expensive (discount / premium)

Cheap and expensive describe price relative to the current interval's algo levels: below the lower levels is cheap, above the upper levels is expensive.

It describes how rare such a deviation was in history, not a call to trade.

More in the video Chart architecture: a chart of random days and why there is no context. See also: Algo level.

Sweep (liquidity sweep)

A sweep is price moving beyond the high or low of a given interval, such as a session or a day.

Studies distinguish the first sweep within the observation window from repeated ones.

More in the video What happens after a session high sweep: 594 of 848 days on US100. See also: Pool, Pullback, Continuation.

Pool

A pool is the high or low of an interval treated as a target for a sweep.

A pool sweep and a level sweep mean the same thing.

More in the video Random concept: from trading concepts to event frequencies. See also: Sweep, Session.

Pullback

A pullback is price moving from the swept level against the sweep direction by a set percentage.

The pullback is awaited within the observation window, for example up to 90 minutes after the sweep.

More in the video Pullback and continuation: why breakout trading loses so often. See also: Sweep, Continuation, Observation window.

Continuation

Continuation is price moving from the swept level further in the sweep direction by a set percentage.

In the pullback-and-continuation study, continuation is looked for after the pullback.

More in the video Pullback and continuation: why breakout trading loses so often. See also: Sweep, Pullback.

Observation window

An observation window is the time after an event, usually 15 to 180 minutes, during which the outcome is recorded: pullback, continuation or neither.

The same term is used for the interval in which the sweep itself is looked for, for example 09:00–15:00.

More in the video What happens after a session high sweep: 594 of 848 days on US100. See also: Pullback, Continuation.

Session

A session is a fixed New York time interval, for example 03:00–04:30 or 07:30–09:30, whose high, low and their sweeps are counted.

Names such as Asia, London or pre-market are conventional labels for the intervals.

More in the video Random concept: from trading concepts to event frequencies. See also: Pool, Trading day.

Trading day

A trading day is the period from 18:00 to 18:00 the next day, New York time.

At 18:00, after the 17:00–18:00 technical break, the US futures evening session opens; all intervals of the day are counted from that moment.

More in the video The math of extremes: how algo levels are computed. See also: Day open, Session.

Day open

The day open is the price at 18:00 New York time, a conditional zero from which deviations are measured in percent.

Deviations can also be measured from the open of the candle itself, the hour or four hours.

More in the video The math of extremes: how algo levels are computed. See also: Trading day, Algo level.

NYM (New York midnight)

NYM is the price at midnight New York time, which some studies use as the reference point for percentages.

See also: Day open.

Data normalization

Data normalization is aligning price series from different sources to one time base and one set of rules so that frequencies are computed on comparable history.

It covers daylight-saving changes, gap checks and reconciliation with exchange data.

More in the video The math of extremes: how algo levels are computed. See also: Sample.

Sample

A sample is the set of days over which a frequency is computed.

It is set by the number of weeks back, months, weekdays and other filters. The fewer days in the sample, the less reliable the frequency: 100% on four cases proves nothing.

More in the video Pullback and continuation: why breakout trading loses so often. See also: Event frequency, Quantile.

Event frequency

Event frequency is the share of sample days on which an event happened under a stated rule.

It is not a forecast and not a win rate: entry, stop and fees are not part of it.

More in the video One rule, two assets: why a universal strategy does not exist. See also: Sample, Sweep.

Replay

Replay is playing back a historical day in which trades can be opened and closed as if the day were happening now.

Future candles are hidden and history plays forward only.

More in the video The BIASMACHINE tester: a tour of the free tester. See also: Tester.

Tester

The tester is the BIASMACHINE section for trading on history: day replay, trades, a journal and prop-challenge emulation.

Free with an account: the tester, deviation quantiles, basic statistics and the CFD 10k and futures-rules 50k challenges. In Pro: date selection, higher speed, fast-forward and extra history, algo levels, blind mode and the advanced journal.

More in the video The BIASMACHINE tester: a tour of the free tester. See also: Replay, Prop-challenge emulation.

Prop-challenge emulation

Prop-challenge emulation is practice on history under typical prop-firm rules: a profit target, a daily loss limit, a maximum drawdown and stages.

No real money is involved and the payout is notional. The CFD 10k and futures-rules 50k challenges are free; other account sizes and rules are in Pro.

More in the video The math of extremes: how algo levels are computed. See also: Tester, Netting.

Netting

Netting is a position mode in which an opposite trade reduces or closes the open position at the average price instead of opening a second one.

More in the video The BIASMACHINE tester: a tour of the free tester. See also: Prop-challenge emulation.

MFE and MAE (maximum favorable / adverse excursion)

MFE and MAE are the largest price moves in the profitable and in the losing direction during a trade.

They are shown in trade analytics (Pro).

More in the video Trade analytics: at which hours and on which attempts you lose. See also: Tester.