← Back to Learn

Taker Intensity Guide

One of Blave's core alpha indicators — what it measures, why it matters, and three ways to build strategies with it.

What is Taker Intensity?

Every trade on a crypto exchange has a maker (the resting limit order) and a taker (the market order that fills against it). Taker Intensity (TI) measures the standardized net difference between market buy orders and market sell orders over a rolling window.

The value is standardized relative to recent history, so it reflects how unusual the current imbalance is compared to the past period — not an absolute level of volume. The same buying volume can produce a TI of +0.5 in a highly active market and +2.5 in a quiet one. Blave provides TI at six timeframes: 15m, 1h, 4h, 8h, 24h, 3d. Longer timeframes are smoother and more suited for position strategies; shorter timeframes are noisier but react faster.

The critical nuance: high TI ≠ always bullish

This is where most people get the indicator wrong. Market buy orders come from two sources:

1. Genuine buyers — traders who want to go long and are willing to pay the ask.
2. Short liquidations — when a short position is forcibly closed, the exchange places a market buy order. This raises TI even though no one is actually bullish.

Three ways to use Taker Intensity

1. Threshold (most common)

Enter long when TI exceeds an entry threshold; exit when TI falls below an exit threshold. The dead zone between the two thresholds prevents the strategy from flip-flopping when TI hovers near a single level.

ENTRY_TH = 1.693
EXIT_TH  = -0.453

def compute_signals(df, entry_th=None, exit_th=None):
    eth = entry_th if entry_th is not None else ENTRY_TH
    xth = exit_th  if exit_th  is not None else EXIT_TH
    ti  = df['TI']
    signal = pd.Series(np.nan, index=df.index)
    signal[ti > eth] = 1.0   # enter long
    signal[ti < xth] = 0.0   # go flat
    return signal

This is exactly the logic in the btc_ti_5min example — TI on a 24h window, checked every 5 minutes. The ENTRY_TH = 1.693 and EXIT_TH = -0.453 values were found by parameter scanning, not guessed.

2. Crossover (simpler, often noisier)

Treat the zero line as the signal threshold. Long when TI crosses above zero, flat when it crosses below. This is intuitive but tends to generate many short-lived trades on 15m or 1h timeframes, running up fees. Works better on 8h or 24h windows where crossovers are more meaningful.

signal[ti > 0] = 1.0
signal[ti < 0] = 0.0

3. Regime filter (most powerful)

Use TI as a gate for another primary signal. Only take entries when TI is positive (net buying pressure exists). This prevents trend strategies from entering into heavy selling environments.

primary = compute_primary_signals(df)  # e.g. SMA cross
ti_positive = (df['TI'] > 0)
signal = primary.copy()
signal[~ti_positive] = 0.0  # suppress entries when TI is negative

Regime filtering typically reduces trade count and fees while improving Sharpe, at the cost of missing some trades. It works best with TI on longer timeframes (8h, 24h) as a macro filter on shorter-interval strategies.

Choosing the right timeframe

In btc_ti_5min, the strategy runs every 5 minutes but uses 24h TI. This is intentional: you want the speed of a 5-minute check (react quickly to TI crossing a threshold) without the noise of a 5-minute TI window.

Combining TI with other indicators

TI in isolation is useful. TI combined with Blave's other alpha indicators is powerful:

← Back to Learn