Module patterns 

Bot-pattern signals in a venue’s recent trades and order book, for the market summary’s health advisory.

Market making is legitimate, so a signal never fails a summary: it warns, with the evidence, where the tape or the book looks machine-made. The signals:

• Repeated sizes: one exact trade size makes up a large share of the tape, or a few sizes make up half of it.
• Metronome timing: trades come at an almost fixed interval.
• Cadence: most gaps between trades are one interval, even when a bot trades in bursts with pauses between them (so the gaps as a whole are not a metronome).
• Ping-pong: a buy and a sell of the same size at nearly the same price, seconds apart (the shape of wash trading).
• Ladder quotes: consecutive book levels at even price steps whose sizes are equal, or rise or fall level by level (a quoting curve).

The functions are pure: they read trades and levels and fetch nothing.

Structs

BotPatterns

The bot-pattern result of one summary.

Enums

Finding

One signal that was found, with its evidence.

LadderShape

How a ladder’s sizes run, from the best level out.

Constants

CADENCE_MIN_GAP_MS

Shortest interval the cadence signal judges: some venues print whole seconds, which would put many short random gaps on one value.

CADENCE_SHARE

Gaps at one interval, as a share of all gaps, at or above this are a cadence. Random arrivals put at most about 1.5% of gaps in any ±2% band.

CADENCE_TOLERANCE

Gaps within this share of an interval count as that interval.

FEW_SIZES

How many sizes the few-sizes signal adds up.

FEW_SIZES_MIN_SHARE

A size counts toward the few-sizes signal only with at least this share of the trades on its own.

FEW_SIZES_SHARE

The FEW_SIZES most common sizes in at least this share of the trades are a signal.

LADDER_DEPTH

How many levels from the top of each side the ladder signal reads.

LADDER_MIN_LEVELS

Fewest consecutive levels that make a ladder.

LADDER_SIZE_TOLERANCE

Levels of a ladder differ in size by at most this share.

LADDER_STEP_TOLERANCE

Price steps of a ladder differ by at most this share of the first step.

METRONOME_MAX_CV

Gaps whose standard deviation is under this share of their mean are a metronome.

MIN_GAPS

Fewest gaps between trade events the timing signal needs.

MIN_PING_PONG_PAIRS

Fewest ping-pong pairs that count as a signal.

MIN_TRADES

Fewest trades the tape signals need. Below this they are not judged.

PING_PONG_PRICE_TOLERANCE_PCT

A ping-pong pair’s prices differ by at most this, in percent.

PING_PONG_SHARE

Trades in ping-pong pairs, as a share of all trades, at or above this are a signal.

PING_PONG_WINDOW_MS

A ping-pong pair is at most this far apart in time.

REPEATED_SIZE_SHARE

One exact size in at least this share of the trades is a signal.

TRADE_SAMPLE

How many recent trades the summary asks the venue for.

Functions

cadence

Cadence: the interval (of at least CADENCE_MIN_GAP_MS) that the most gaps between trade events are at, within CADENCE_TOLERANCE, when those gaps are at least CADENCE_SHARE of all gaps.

detect

Runs every signal. trades is the venue’s recent tape, or why there is none. book is the bids and asks of a real exchange book, or None for a synthetic one (a DEX pool’s curve is an even ladder by design).

few_sizes

Few sizes: the FEW_SIZES most common sizes (each with at least FEW_SIZES_MIN_SHARE) cover FEW_SIZES_SHARE of the trades. A bot that rotates through a few sizes passes under repeated_size.

ladders

Ladder quotes on either side of a real exchange book.

metronome

Metronome timing: the whole tape, else the dominant size’s trades, at a near-fixed interval.

ping_pong

Ping-pong: buy/sell pairs of the same size within PING_PONG_WINDOW_MS and PING_PONG_PRICE_TOLERANCE_PCT. Each trade is in one pair at most. Returns the finding and the paired trades.

repeated_size

Repeated sizes: the dominant size and the trades it covers, when it reaches REPEATED_SIZE_SHARE.