MF
MarketForecastBeta
← Back to home

Forecast Accuracy

Every forecast published on this site is written to an append-only ledger the moment it is generated, then scored against what actually happened once its 30-day horizon elapses. This page is that record.

What is being measured

We forecast a range with a probability, not a single number. So the question is not "was the price right" — it is calibration: when we say 80%, does it happen 80% of the time?

The central estimate deliberately sits close to the current price. Over a 30-day horizon a random walk is very hard to beat, and models that appear to beat it usually do so by extrapolating recent trends, which measurably increases error. We report our point error next to a random walk's so you can see we are matching it rather than claiming to beat it.

80% band coverage

54.9%

target 80.0%

50% band coverage

31.6%

target 50.0%

Median error

15.99%

random walk 15.70%

Resolved forecasts

942

1229 pending

Reliability

Forecasts grouped by the probability we stated. Perfect calibration means observed frequency equals stated probability in every bucket.

5% stated
18% actualn=539
15% stated
25% actualn=432
26% stated
40% actualn=550
32% stated
9% actualn=363
47% stated
51% actualn=53
56% stated
52% actualn=240
65% stated
47% actualn=522
73% stated
67% actualn=118
81% stated
89% actualn=9

The white tick marks where the bar would end under perfect calibration.

By asset

Assetn80% band50% bandMed. errorvs walkBrier
xrp234.3%4.3%25.18%24.60%1.311
arbitrum238.7%8.7%47.44%47.04%1.079
avalanche2369.6%8.7%16.91%15.98%0.727
bitcoin238.7%4.3%18.24%18.12%0.938
bnb238.7%0.0%16.93%16.36%0.839
aptos2387.0%69.6%7.20%7.36%0.445
cardano2391.3%47.8%12.77%11.64%0.521
chainlink2326.1%4.3%26.15%25.83%0.887
copper23100.0%95.7%1.85%2.33%0.299
corn2334.8%8.7%11.46%11.45%1.102
cosmos2373.9%21.7%15.02%13.92%0.129
dogecoin2313.0%4.3%17.26%16.58%0.564
ethereum2313.0%8.7%23.53%23.57%0.883
ethereum classic2326.1%0.0%17.61%16.52%0.882
filecoin2369.6%30.4%14.65%13.73%0.666
gold2382.6%73.9%1.92%2.07%0.314
hedera2347.8%8.7%12.79%12.01%0.334
injective2352.2%34.8%22.57%23.14%0.768
internet computer2330.4%0.0%17.89%17.58%0.858
litecoin2321.7%0.0%15.97%15.65%0.934
mantle2313.0%0.0%28.16%26.58%0.866
monero238.7%0.0%27.96%28.33%0.922
naturalgas2387.0%87.0%3.78%3.10%0.447
near2326.1%21.7%29.70%30.41%0.747
oil2387.0%65.2%12.08%12.84%0.344
optimism2378.3%26.1%14.19%12.46%0.495
palladium23100.0%91.3%2.04%2.03%0.296
platinum23100.0%82.6%4.19%4.43%0.166
polkadot2343.5%13.0%21.63%20.70%0.809
render2391.3%30.4%10.40%10.23%0.217
shiba inu23100.0%39.1%12.35%11.44%0.463
silver23100.0%78.3%3.40%3.88%0.115
solana234.3%0.0%26.91%26.60%1.189
stellar2382.6%34.8%10.72%10.67%0.189
sugar2360.9%26.1%6.96%7.23%0.826
sui2382.6%43.5%9.81%9.22%0.485
tezos2343.5%21.7%20.69%19.24%0.677
tron23100.0%100.0%1.31%1.62%0.387
uniswap238.7%0.0%42.25%42.49%1.082
wheat2378.3%30.4%9.51%9.74%0.858
brent2286.4%72.7%14.06%14.84%0.411

How to read this honestly

  • Small samples mean little. A coverage figure over a handful of resolved forecasts is noise; treat anything under ~30 samples per asset as indicative only.
  • Overlapping horizons are correlated. Consecutive daily forecasts share most of their 30-day window, so the effective sample size is well below the raw count.
  • Good calibration in a calm market says little about a violent one. Volatility estimates lag regime changes, and the bands will be too narrow until they catch up.
  • None of this makes any individual forecast reliable. A well-calibrated 70% is still wrong 30% of the time, and that is the point of publishing the number rather than a headline price.

Methodology and model details are on the methodology page. Forecasts are logged daily and scored automatically; nothing on this page is entered by hand.