mylittleindicators

Multi-stream financial indicators library for Rust. 559 bar indicators + 21 event primitives across 35 categories. Consumes the full surface of digdigdig3 exchange connectors — OHLCV bars, ticks, orderbook (snapshot + delta), funding, mark price, open interest, liquidations, ticker, agg trades, plus 12+ extended stream types (option greeks, basis, index price, settlement, block trades, L3, etc.).

What's inside

bar_indicators/ — 559 single-stream indicators (averages, momentum, volatility, channels, divergence, kalman, signal processing, statistics, candles, levels, accumulation, adaptive, regression, chaos, entropy, trend, trend-stops, zigzag, ratio, position, statistical scoring, sentiment, ticker, mark price, funding, OI, liquidations, book, book advanced, clusters, hybrid tick+book, microstructure, tick advanced, greeks, index basis, volatility advanced, stress, risk funding, composites).
events/ — 21 event primitives (swing detection, divergence, candle patterns, line cross, price-line cross, pivot, threshold, volume events, volatility regime, regime gate, relative position, statistical wick, confluence, BOS, FVG, oscillator+divergence, oscillator+volume weight, direction detector, and 3 cross-asset events: beta, cointegration proxy, relative strength).
catalog/ — IndicatorSignature + EventSignature with input_stream / aux_streams routing metadata. Used by codegen, UI rendering, live validators.
data_loader/ — StreamKind, TimedEvent, EnrichedHistory, RestFetcher abstraction over dig3.

Two parallel factories

use mylittleindicators::bar_indicators::{
    bar_indicator_id::BarIndicatorId,
    instance_factory::{IndicatorConfig, IndicatorInstance},
};

let cfg = IndicatorConfig::new(BarIndicatorId::Rsi, "rsi".into(), vec![14]);
let mut rsi = IndicatorInstance::create(&cfg)?;
let v = rsi.update_bar(open, high, low, close, volume, Some(ts_ms));

use mylittleindicators::events::{EventId, EventConfig, EventInstance};

let cfg = EventConfig::new(EventId::SwingDetection, "swing".into())
    .with_string_param("mode", "percent")
    .with_param("threshold_pct", 0.5);
let mut swing = EventInstance::create(&cfg)?;
let v = swing.update_bar(o, h, l, c, vol);

Both factories follow the same shape: Id enum + Config builder + Instance::create() + update_* dispatch + value() / is_ready() / reset().

Stream coverage

Every IndicatorSignature / EventSignature declares which streams it consumes:

pub struct IndicatorSignature {
    pub input_stream: StreamKind,            // primary stream
    pub aux_streams: &'static [StreamKind],  // secondary streams (multi-stream composites)
    // ... + role_kind, output_kind, params, ...
}

StreamKind mirrors dig3 exactly (Bar, Tick, OrderBook, OrderbookDelta, Funding, MarkPrice, OpenInterest, Liquidation, Ticker, AggTrade, LongShortRatio, OptionGreeks, VolatilityIndex, HistoricalVolatility, Basis, IndexPrice, CompositeIndex, InsuranceFund, Settlement, BlockTrade, OrderbookL3, RiskLimit, PredictedFunding, FundingSettlement, MarketWarning, MarkPriceKline, IndexPriceKline, PremiumIndexKline).

Consumers route events by checking signature.accepts(kind) == primary == kind || aux_streams.contains(&kind).

OrderBook DOM pattern

For live consumers (chart apps, backtesters): mirror mylittlechart::orderbook-service — maintain a dig3-station::OrderBookTracker per (exchange, symbol). On every event:

Event::OrderbookSnapshot { point, .. } => {
    tracker.apply_snapshot(&book);
    // feed snapshot to indicators
}
Event::OrderbookDelta { point, .. } => {
    tracker.apply_delta(&delta);
    // reconstruct full book from tracker.top_bids(50) / top_asks(50)
    // feed reconstructed book to snapshot-aware indicators
}

This is what real-world UIs do — 1 snapshot from the exchange + 1000+ delta updates per minute = continuous DOM state for all book-aware indicators.

Hard rules

BarIndicatorId is single-stream. Multi-symbol (cross-asset beta) and multi-TF indicators live in events::.
No hardcoded magic numbers in IndicatorInstance::create(). All thresholds/multipliers/periods come from config.additional_params with sensible unwrap_or defaults — so runtime tuning by validators and UIs works.
is_ready() must return true after any valid update path (bar, tick, or L2), not just the original one.
Every BarIndicatorId / EventId MUST have a catalog signature. Discovery via MasterCatalog::iter_signatures() / MasterEventCatalog::iter_signatures().

Status

Verified live against digdigdig3 v0.3.10 on 12 exchanges (Binance, Bybit, OKX, Bitget, GateIO, HTX, HyperLiquid, KuCoin, MEXC, Deribit, BitMEX, Bitstamp) covering all 27 dig3 Stream variants — core market data (Trade/AggTrade/Bar/Ticker/Orderbook/OrderbookDelta) on Binance + Bybit + Deribit, derivative data (Funding/Mark/OI/Liquidation) on 7 venues, extended streams (LongShortRatio polling, HistoricalVolatility polling, OrderbookL3, Basis derived, FundingSettlement derived, PredictedFunding from BitMEX + OKX, OptionGreeks/VolatilityIndex on Deribit, MarkPrice/IndexPrice klines on OKX/Binance/GateIO).

Pass rate on a 150s live BTC slice: ~89% combined (494 / 556 bar indicators, 21 / 21 events = 100%, 514 / 577 total). Newly passing in 0.1.2: LongShortRatioMomentum, HvMomentum, L3OrderRate, L3LargeOrderTracker, L3SpooferScore, RegimeGate.

Remaining failures are environment-bound, not code: streams that need actual market events to fire (liquidation cascades, settlement-window 8h transitions, option-flow surges) or that the tested venues don't emit on this instrument-set (CompositeIndex is Binance index-symbol- specific; MarketWarning/RiskLimit/InsuranceFund are broadcast only on real fund-change / margin events).

The validator binary is included in crates/mli-validator/ (dev-only diagnostic harness, publish = false) — run it yourself to see the pass/never_ready/always_zero/never_received_event matrix:

cd crates/mli-validator

cargo build --release --bin mli-validator

./target/release/mli-validator --duration-secs 60

cat validator_report.json

License

MIT

mylittleindicators 0.1.8