Timeframe (Prediction Horizon)
How far into the future a prediction looks. LDBD offers 1d, 1w, and 1m today (6m and 1y are on the way) and annualizes each so they compete on one leaderboard.
A timeframe is how far out a single prediction looks. On LDBD you pick one of the live horizons — one day (1d), one week (1w), or one month (1m); six months (6m) and a year (1y) are on the way — and then call the direction, up or down, over that span.
So how do calls with wildly different horizons share one leaderboard? Each timeframe carries a fixed holding period (1d is 1/252 of a year, 1y is a full year), and returns are annualized using that number, so a one-day call and a one-year call land on the same yardstick. Skill Rating leans the other way — shorter horizons get a bigger K-factor, running from 32 at 1d down to 8 at 1y. The two long horizons, 6m and 1y, are also capped at one prediction per asset per week.
Example
Holding periods: 1d = 1/252 yr, 1w = 5/252, 1m = 21/252, 6m = 126/252, 1y = 1. Skill Rating K-factor: 1d = 32, 1w = 24, 1m = 16, 6m = 12, 1y = 8.
How LDBD uses it
Every prediction commits to a timeframe up front, and it shapes the scoring that follows. The headline rate divides each return by that timeframe's holding period to annualize it, which is why a 1d and a 1y bet can be ranked together. The Skill Rating K-factor also varies by timeframe (32 at 1d down to 8 at 1y).
FAQ
How can predictions with different timeframes be compared fairly?
Each return is divided by its timeframe's holding period to put it on a one-year basis, then averaged. Whether the gain came in a day or over a year, it's restated as 'what it would have been across a full year' and set side by side.