Aperiodic
Metrics
MarketsLive
DocsPricing
Get the data

Metrics library · Order flow

Flow indices

Sample CSV
BTC-USDT perpetual on Binance, 5m bars, 1 to 31 May 2025, from the free preview. The free delayed feed doesn't carry this dataset; paid plans stream it in real time.
How it's builtBTC, May 2025FieldsDownloadFAQ

Conviction is the share of recent bars one side of the takers won. Flow pressure is who won this one.

Two bounded reads of taker flow: conviction, the time-decayed share of recent bars one side won, and flow pressure, the net aggressor share of the bar.

Built from
Flow Metrics
Fields
2
Intervals
15m to 1d
Plan
Prime
Venues
Binance, OKX, Hyperliquid
Live
Real time on paid plans

How it's built

One real 5-minute bar from the free preview, field by field.

The day of 5m bars up to the bar from 14:00 to 14:05 UTC on 14 May 2025, Binance BTC-USDT perpetual, from the free preview.

  1. Step 1

    Who won the bar

    flow_pressure=100 × (taker buy − taker sell) ÷ (taker buy + taker sell)=−5.98

    Sellers took slightly more of this bar's notional. The figure swings bar to bar: it says who won one bar, not who has been winning.

    +58−63
    14:05, 13 May14:00, 14 May

    flow_pressure on each 5-minute bar of the day: above 0 buyers took more notional, below 0 sellers did.

  2. Step 2

    A vote per bar

    vote=sign(taker buy − taker sell)=−1

    Over the day, buyers won 139 bars and sellers 149. Each bar counts once however much traded, so one block trade can't carry the day.

    +1−1
    08:0514:00

    The last 6 hours of votes: +1 for a bar buyers won, −1 for one sellers won.

  3. Step 3

    Average the votes with a 6-hour decay

    conviction=Σ vote × e^(−age / 6h) ÷ Σ e^(−age / 6h)=−0.060

    Close to 0: over the last hours the wins were split almost evenly. Near ±1 one side would have won nearly every bar.

    +0.09−0.22
    14:05, 13 May14:00, 14 May

    conviction through each bar. A vote 6 hours old counts e⁻¹ ≈ 37% as much as this bar's.

  4. Step 4

    The bar

    conviction
    −0.0600
    flow_pressure
    −5.976

What BTC did after a run of one-sided buying

In May, BTC fell 5.75 bps on average in the hour after its most one-sided buying, and rose in every other group.

BTC-USDT perpetual on Binance, 5m bars, 1 to 31 May 2025, from the free preview.

Mean BTC return over the next hour
2.53
6.48
2.40
1.74
-5.75
Most sellingSellingEvenBuyingMost buying

Bars by conviction, in five equal groups. Basis points.

Sorted into five equal groups by conviction, the bars with the most persistent buying (conviction above 0.045) were followed by a mean return of -5.75 bps over the next hour, and 47% of them by a rise. The other four groups returned 1.74 bps to 6.48 bps, 53% of them up.

Conviction moves slowly: bars an hour apart correlate 0.83, against 0.02 for flow pressure, which tracks the bar's own return (0.58). One month of one contract is a hypothesis, not a signal. Test it on the history before you trade it.

How to read it

flow_pressure is 100 × (taker buy − taker sell) ÷ (taker buy + taker sell) notional in the bar. +100 means every taker bought, −100 that every taker sold. It says who won this bar, so it moves with the bar's own return.

conviction gives each bar a vote of +1 if buyers took more notional, −1 if sellers did, and averages the votes with a 6-hour time decay, the current bar included. Near ±1, one side has won almost every recent bar; near 0, wins are split. It ignores how big each bar was, so one huge print can't make it.

Both are computed per venue and contract. conviction is NULL until the contract has 10 earlier bars, the first a week or more back.

conviction
Time-decayed mean, with a 6-hour decay through the current bar, of the sign of taker buy minus taker sell notional, from −1 to 1. +1 when buyers won every recent bar, −1 when sellers did; NULL until 10 earlier bars, the first a week or more back
flow_pressure
100 × (taker buy − taker sell) ÷ (taker buy + taker sell) notional in the bar, from −100 (every taker sold) to 100 (every taker bought)

Download the same month

No account needed. Run one command, or hand it to your agent.
# Install the Aperiodic CLI into ~/.local/bin
mkdir -p "$HOME/.local/bin"
curl -fsSL https://raw.githubusercontent.com/aperiodic-io/cli/main/install.sh | INSTALL_DIR="$HOME/.local/bin" bash
export PATH="$HOME/.local/bin:$PATH"

# Free preview — no API key required
aperiodic flow_indices --preview \
  --exchange binance-futures \
  --symbol perpetual-BTC-USDT:USDT \
  --interval 5m \
  --timestamp exchange \
  --start-date 2025-05-01 \
  --end-date 2025-05-31 \
  --output-dir ./data && echo "Saved to $PWD/data"

Every venue and perp, from 1-minute to daily bars, comes with Prime.

Flow indices in the API docs: its endpoint and parameters →

Questions

Related

FlowTaker buys against sells, by size, with toxicityPositioningOpen interest building, against price, volume and fundingTrade sizeSmall, medium and large trades, and the tails
Aperiodic

Built from hundreds of terabytes of raw data and distilled into point-in-time metrics you can pull as parquet files. Plus the raw trades, quotes and derivatives data behind them.

Data
  • Metrics library
  • Order Flow
  • L1 & L2
  • Derivatives
  • Raw Data
  • Data Catalog
  • Free data
  • Pricing
Use Cases
  • Improve your backtest
  • Find new alpha
  • Execute smarter
Indices
  • All indices
  • Flow indices
  • Pressure Index
  • Conviction Index
  • Positioning indices
  • OI Divergence Index
  • Leverage Index
  • Crowding Index
  • Liquidity indices
  • Cost Index
  • Depth Index
  • Impact Index
Tokenized markets
  • All real-world assets
  • Commodity perps
  • Tokenized stock perps
  • Stock index perps
  • Tokenized ETF perps
  • Pre-IPO perps
  • FX perps
  • Hyperliquid HIP-3
Live markets
  • Live markets dashboard
  • Every perp, A–Z
  • Real-world asset board
  • Global Liquidity
  • Unusual activity
  • Live funding rates
Developers
  • Docs
  • API Docs
  • MCP Server
  • For AI Agents
  • Pre-flight
  • Notebooks
  • Integrations
  • Atlas Security Master
  • TParquet
  • llms.txt
Learn
  • Markets from First Principles
  • Microstructure Guide
  • Replays
  • Architecture
  • Research Blog
  • FAQ
Company
  • Team
  • Contact
  • Book a call
  • Changelog
  • Roadmap
Compare
  • All vendors
  • Pricing compared
  • vs. Tardis
  • vs. CoinGlass
  • vs. Kaiko
  • vs. Amberdata
  • vs. Crypto Lake
  • vs. Laevitas
  • vs. Coinalyze
Exchange data
  • Binance archive
  • Hyperliquid data
  • Binance OI history API
  • OKX history APIs

© Copyright 2026 Aperiodic. All Rights Reserved.

Aperiodic Limited · Registered in Ireland · Company No. 815273 ·
136 Capel Street, Dublin D01 T2C9, Ireland
Terms of ServicePrivacy PolicyLicensingSLA