Skip to main content

Evaluate a strategy against your book

The question you actually have: does adding a BTC Alpha strategy to my portfolio improve its risk-adjusted return? Two snippets below — a no-key taste you can run right now, and the full recipe that measures a strategy against your returns.

:::note Factual data, not advice Everything here is backtested / modelled historical data, provisional pending independent (TradingView) verification. Past performance is not indicative of future results — not advice, not a recommendation, not a forecast. The example numbers are illustrative; run it on your own data. :::

Taste it now — no key​

How independent are the 14 strategies from each other? One open call:

import requests
API = "https://btc-strategy-data-api.fly.dev"
s = requests.get(f"{API}/v1/summary").json()["data"]          # free — no key
lc = s["diversification"]["least_correlated"]
print(f"{s['strategy_count']} strategies, mean pairwise correlation {s['mean_pairwise_correlation']:.2f}")
print(f"-> behave like {s['diversification']['effective_bets']:.1f} independent bets")
print(f"least-correlated pair: {lc['pair'][0]}/{lc['pair'][1]} at {lc['correlation']:+.2f}")
14 strategies, mean pairwise correlation 0.35
-> behave like 2.5 independent bets
least-correlated pair: helios/kronos at -0.09

What to notice: a pair at −0.09 is negatively correlated — a hedging return stream, not just a diversifier. Low mutual correlation is the raw material; the step that matters is how a strategy sits against your book — that's the recipe below. (Backtested/modelled, provisional — not advice.)

The full recipe — vs your book (free, no key)​

Pull a strategy's full daily return series (free — no key), align it with your own daily returns, and see the correlation and the effect on your Sharpe.

# pip install requests numpy
import csv, requests, numpy as np

API   = "https://btc-strategy-data-api.fly.dev"
STRAT = "helios"                 # any slug from GET /v1/strategies

# 1 — the strategy's full daily return series (free — no key)
r = requests.get(f"{API}/v1/strategies/{STRAT}/returns",
                 params={"period": "daily"})
r.raise_for_status()
strat = {p["t"]: p["ret"] for p in r.json()["data"]}

# 2 — YOUR book's daily returns, keyed by the same ISO dates (see "Plug in your returns" below)
mine = {row["date"]: float(row["return"]) for row in csv.DictReader(open("my_returns.csv"))}

# 3 — align on common dates, then measure
dates = sorted(set(strat) & set(mine))
s = np.array([strat[d] for d in dates])
b = np.array([mine[d]  for d in dates])
sharpe = lambda x: x.mean() / x.std(ddof=1) * np.sqrt(365)

rho = np.corrcoef(s, b)[0, 1]
print(f"correlation to your book : {rho:+.2f}   (near 0 = independent return stream)")
print(f"your book Sharpe         : {sharpe(b):.2f}")
for w in (0.10, 0.20, 0.30):
    print(f"  + {w:.0%} {STRAT:<7} -> Sharpe {sharpe((1 - w) * b + w * s):.2f}")

Illustrative run — here "your book" is a sample diversified book (an equal-weight blend of the other strategies), shown only so you can see the shape of the answer:

correlation to your book : +0.16     (near 0 = an almost independent return stream)
your book Sharpe         : 1.46
  + 10% helios -> Sharpe 1.67
  + 20% helios -> Sharpe 1.74
  + 30% helios -> Sharpe 1.69

:::warning Illustrative — not a promise of your result That sample book is not your portfolio, and these are not projected outcomes. The real answer is whatever you get with your returns plugged in. Backtested/modelled, provisional, not advice. :::

A near-zero correlation means the strategy is a largely independent return stream, so a modest sleeve can raise a book's risk-adjusted return through diversification — if it holds for yours.

Plug in your returns​

mine is a {ISO-date: daily-return} dict. The snippet reads a CSV with two columns:

date,return
2018-08-15,0.0042
2018-08-16,-0.0011
...

Use decimal daily returns (0.0042 = +0.42%), aligned to UTC dates. Any source works — swap the CSV loader for your dataframe.

Variations​

  • Whole bundle, not one strategy. Loop the call over several slugs (e.g. a bundle's members) and blend the equal-weight sleeve into your book.
  • Best sleeve size. Sweep w on a finer grid and pick the weight that maximises sharpe.
  • Pick the least-correlated first. Use GET /v1/correlation (free, no key) to rank strategies by how independent they are.

Verified it? Subscribe for the live signals​

Everything above is free — verify every track record against your own book before you commit to anything. When you want to act on these strategies in real time, the product is the live signals: actionable trade signals as each strategy fires (is_test: false), delivered by the keyed signals API.

Live signals are request-based today — contact for access.