Know the phase. Time the turn.
Phase forecasts, turn dates and honest backtests for crypto, stocks and metals.
Private beta, plans coming soon.
One verdict per asset, scored in the open
Phase, turn date, confidence
Growth or decline, the next likely turn date and a grade of High, Moderate or Uncertain, refreshed as new bars close.
Honest backtests
Every forecast is stored and re-scored walk-forward. Hit rate, average error in days and max drawdown are shown as they are.
Trade ideas
When the timeframes align, the model drafts an entry zone, a stop and targets, then tracks whether each was hit.
How the model works
Trend, momentum and cycle detection run on 1H, 4H and 1D bars. Each timeframe votes; the verdict is their weighted agreement.
- 01
Trend
Moving averages at 20, 50 and 200 periods, measured against volatility.
- 02
Momentum
RSI, MACD and rate of change. Exhaustion pulls the turn date earlier.
- 03
Cycles
Swing-to-swing distances and autocorrelation give the dominant period and where you are in it.
- 04
Agreement
The three timeframes vote. Full agreement earns a High grade and the trade filter.
Honest by design
A forecast is written to the database the moment it is made and never edited. Walk-forward backtests re-run the model bar by bar using only the history it could have seen, then mark each call with hindsight: hit or miss, error in days, and the drawdown of simply following the phase.
The app shows the live numbers with their sample counts, next to the disclaimer. This page shows none.
- hit rate
- live in app
- error, days
- live in app
- max drawdown
- live in app
News and macro, in the same screen
Crypto and US market headlines, FRED macro series and the economic calendar sit beside the chart, so a scheduled CPI print or FOMC decision is in view before you act on a turn.
Explain it to me
Ask why an asset got its verdict. The Claude-powered explainer reads the stored forecast and its inputs, trend, momentum, cycle position and timeframe agreement, and answers in plain language.
It explains the model’s verdicts. It never makes predictions of its own.
The daily and 4-hour timeframes agree on growth: price sits above its 50-period average and momentum is positive. The 1-hour timeframe has turned down, so the three do not fully agree, which lowers the grade from High to Moderate.
The turn estimate comes from the median length of the last eight swings on the daily chart, pulled earlier because RSI is stretched.
Explains the stored verdict and its inputs. Makes no predictions of its own.
Coverage
Nine assets, three timeframes each: 1H, 4H and 1D.
- BTCBitcoinCrypto
- ETHEtherCrypto
- SOLSolanaCrypto
- NVDANvidiaUS stocks
- TSLATeslaUS stocks
- AAPLAppleUS stocks
- SPXS&P 500Index
- XAUGoldMetals
- XAGSilverMetals