DSL Reference

Complete reference for the .algomln strategy language.

AlgoMLN strategies are written in .algomln files — a small, intentionally limited domain-specific language. It is rules-only: no variables, no loops, no state. Just conditions and actions. The simplicity is by design — strategies should be readable at a glance.

The language is case-insensitive for keywords. Blank lines and # comments are allowed anywhere. Indicator periods and quantities must be positive integers.

Grammar

Formal grammar of the .algomln language in EBNF notation:

strategy       = rule+

rule           = "WHEN" condition NEWLINE action

condition      = comparison
               | cross_expr
               | not_expr
               | logical_expr
               | position_expr    ← parses, not yet evaluated
               | time_window      ← parses, not yet evaluated

comparison     = expr operator expr
operator       = "<" | ">" | "<=" | ">=" | "==" | "!="

logical_expr   = condition "AND" condition
               | condition "OR" condition

not_expr       = "NOT" "(" condition ")"

cross_expr     = "cross_above" "(" expr "," expr ")"
               | "cross_below" "(" expr "," expr ")"

expr           = indicator_call | price_field | number

indicator_call = indicator "(" integer ")"
indicator      = "ema" | "ma" | "rsi" | "rel_vol" | "atr" | "vwap"
               | "bb_upper" | "bb_lower" | "bb_mid"

price_field    = "close" | "open" | "high" | "low" | "volume"
               | "prev_close" | "prev_open" | "prev_high" | "prev_low"

action         = "BUY" integer
               | "SELL" integer
               | "SELL" "ALL"

Price Fields

These identifiers resolve to values from the current (or previous) candle:

Indicators

All built-in indicators are pure Rust functions — stateless and deterministic. They use a bounded window provider to stay O(N) regardless of history length. Custom indicators can be registered via the Plugin System.

Indicator calls take the form indicator_name(period) where period must be a positive integer.

ma — Simple Moving Average

Arithmetic mean of the last N close prices.

WHEN close > ma(50)
BUY 10

ema — Exponential Moving Average

Exponentially weighted moving average, giving more weight to recent candles. Used for crossover strategies.

WHEN cross_above(ema(20), ema(50))
BUY 10

rsi — Relative Strength Index

Momentum oscillator in the range 0–100. Values below 30 indicate oversold; above 70 indicate overbought. Typical period: 14.

WHEN rsi(14) < 30
BUY 1

WHEN rsi(14) > 70
SELL ALL

atr — Average True Range

Measures market volatility. The true range is the greatest of: current high−low, |current high−prev close|, |current low−prev close|. ATR averages this over N candles.

WHEN atr(14) > 50
BUY 5

vwap — Volume Weighted Average Price

Average price weighted by volume over the given period. Commonly used as a dynamic support/resistance level.

WHEN close > vwap(20)
BUY 5

rel_vol — Relative Volume

Current candle volume relative to the average volume over N candles. A value of 2.0 means twice the average volume — useful for detecting breakouts.

WHEN rel_vol(20) > 2 AND close > ma(50)
BUY 10

bb_upper / bb_mid / bb_lower — Bollinger Bands

Bollinger Bands consist of a middle band (simple moving average) and upper/lower bands at ±2 standard deviations. All three share the same period parameter and are computed together for efficiency.

WHEN close < bb_lower(20)
BUY 10

WHEN close > bb_upper(20)
SELL ALL

Conditions

Conditions follow the WHEN keyword and determine when a rule fires.

Comparison

Compare any two expressions using a standard operator:

WHEN close > ma(50)
BUY 5

WHEN rsi(14) >= 70
SELL ALL

Cross Detection

Crossovers fire on the exact candle where the fast value crosses the slow value. The CrossDetector internally stores the previous candle values for each rule and compares them after evaluation — so all rules within a cycle see consistent previous-candle state.

# Fires when ema(20) crosses above ema(50)
WHEN cross_above(ema(20), ema(50))
BUY 10

# Fires when ema(20) crosses below ema(50)
WHEN cross_below(ema(20), ema(50))
SELL ALL

Logical — AND / OR / NOT

Combine conditions with AND, OR, or negate with NOT(...).

# AND — both must be true
WHEN ema(9) > ema(21) AND rsi(14) < 60
BUY 5

# OR — either is sufficient
WHEN rsi(14) > 75 OR close > bb_upper(20)
SELL ALL

# NOT — negation
WHEN NOT (rsi(14) > 70)
BUY 2

Actions

Actions follow the condition on the next line. They instruct the execution engine to place an order.

Quantities (n) must be positive integers. Partial sells greater than held units are clamped to the position size by the PaperBroker.

Examples

RSI Oversold / Overbought

WHEN rsi(14) < 30
BUY 1

WHEN rsi(14) > 70
SELL ALL

EMA Crossover

WHEN cross_above(ema(20), ema(50))
BUY 10

WHEN cross_below(ema(20), ema(50))
SELL ALL

Compound Condition

WHEN ema(9) > ema(21) AND rsi(14) < 60
BUY 5

WHEN rsi(14) > 75
SELL ALL

Bollinger Band Breakout

WHEN close < bb_lower(20)
BUY 10

WHEN close > bb_upper(20)
SELL ALL

VWAP + Volume Filter

# Only buy if volume is elevated and price is above VWAP
WHEN close > vwap(20) AND rel_vol(20) > 1

BUY 5

WHEN close < vwap(20)
SELL ALL

Limitations

These features parse successfully but are not yet evaluated at runtime. Using them will not cause an error, but they will have no effect:

Additional intentional constraints:

• No variables — expressions cannot be assigned to names
• No loops — strategies run one candle at a time, not imperative loops
• Indicator periods must be positive integers — no expressions or variables as periods
• No string literals — the DSL has no string type