Deptho Pro
- Test every variant at tick level, not on candles.
- Months of ticks on your laptop, at no data cost.
- Works with orders from any backtester or pandas.
Under a 512 MB cap, Deptho Pro replayed 10,000 orders across 90M trades in 1.4 seconds, where pandas, Polars and DuckDB ran out of memory.
Fills are modeled from trade prints; order book depth is coming with BookBuilder.
Draft pricing, not final.
Pro Individual
$24.17/mo billed yearly, $58 less than monthly. One key per year instead of one per month. 14-day refund window.
One person trading their own capital or a client's. Commercial use.
Pro Team
Minimum 3 seats ($597/month).
Crypto funds and desks. Seats are counted per person.
Academic
Pro Individual for non-commercial research and teaching.
Open Source
The deptho library: perp tearsheets with funding and fee attribution, plus the demo. Apache-2.0.
Early buyers keep their price, including when order book replay (BookBuilder) ships.
What you get
dpp.sync(symbol, start, end): daily Binance archive files kept as one append-only dataset. It repairs itself after an interrupted sync and rebuilds if the inputs change order.- Trade-print execution replay across the whole dataset, in parallel. Same fills as the pandas, Polars and DuckDB implementations in the benchmark, and closer to exact arithmetic.
- 1-second (or any) OHLCV/VWAP bars and rolling statistics over the full tick history
dp.loadroutes to the engine automatically once a license is active- Linux, macOS (Intel and Apple Silicon) and Windows wheels
Benchmark
warm / cold · no cap · Methodology
| Engine | Warm | Cold | Peak heap | Peak RSS | Lines |
|---|---|---|---|---|---|
| pandas + Parquet | 13 s | 17 s | 7.4 GB | 7.5 GB | 17 |
| Polars + Parquet | 13 s | 23 s | 9.1 GB | 9.2 GB | 20 |
| DuckDB | 119 s | 117 s | 11.6 GB | 11.7 GB | 21 |
| Deptho Pro | 0.10 s | 1.13 s | 70 MB | 1.4 GB | 3 |
Replay of 10,000 market orders across 90 days / 90,000,000 Binance-format trades (trade-print fill model), each engine on its own idiomatic storage; every engine produces the same fills. Warm: files in page cache. Cold: page cache dropped. Peak heap is memory the process allocates for itself; peak RSS also counts memory-mapped file pages, which for Pro is mostly the dataset cached by the OS (evictable whenever it needs the space). Intel(R) Xeon(R) Processor @ 2.10GHz, 4 CPUs, 16 GB RAM, Ubuntu 24.04.5 LTS.
How activation works
- Subscribe. Your license key appears on the next page and is also emailed to you.
pip install deptho-pro --index-url "https://<YOUR_KEY>@pkg.deptho.io/simple/"python -c "import deptho_pro; deptho_pro.activate('<YOUR_KEY>')"
FAQ
- Does Pro send my data or strategy anywhere?
- No. The license key is checked offline against a public key built into the package, so Pro works on air-gapped machines. The only network call either package makes is a data download you start yourself, and it goes to the exchange's public archive.
- Why is the yearly plan featured?
- Keys are valid for the paid period plus 7 days. On a monthly plan you activate a new key each month (it's emailed to you); on a yearly plan, once a year.
- What happens if I cancel?
- Your key keeps working until the end of the paid period plus 7 days. After that,
dp.loadfalls back to the free pandas path. Datasets you've already synced stay on your disk. - Can I get a refund?
- Yearly plans can be refunded in full within 14 days of purchase. Monthly plans aren't refunded: cancel any time, and your license continues through the period you've paid for. See the terms.
- What happens if I'm refunded?
- The license ends, and the package index stops accepting your key straight away.
- What does replay fill against?
- Trade prints: the VWAP of the trades that followed your order on the same side. It doesn't model order-book depth or queue position yet.
- Isn't this what Polars and DuckDB do?
- For many jobs, yes, and they're excellent. Polars ingests CSV faster than Pro and stores it smaller (see the full benchmark). Pro's advantage is replay across the whole dataset without loading it into RAM.