Learned projection
The learned projection draws a forecast line that a small neural network learns from inputs you choose: the Moon's phase, planetary angles and places, the annual cycle, Bitcoin's halving cycle, or the cycle projection's own cycles.
Cycles studyDaily bars
What it's for
- Testing whether a set of inputs carries any forecast at all.
- Comparing a learned line with a straight fit on the same inputs.
How to turn it on
- 1Open Studio on the market you want, on a daily chart (it runs on daily bars).
- 2Open Studies in the top bar → Cycles → Learned projection.
- 3Or open the Models panel → Learned model, from + Studies or the inspector.
How to read it
Choosing the inputs is the whole craft. The panel then shows what the line leans on, and, when asked, how it did walk-forward and beside a straight fit.
The steps are numbered: show the line, score it, measure it against noise, share it. A score means little until it has been set against noise.
Its last step freezes the settings and lets the engine forecast with them, sealed, window after window, so the record builds on bars no one has seen.
Settings worth knowing
- Inputs
- Moon phase; Jupiter–Saturn, Sun–Mars and Sun–Venus angles; Jupiter's and Saturn's places; the annual cycle; the halving cycle (Bitcoin only); the fitted cycles, refitted each step. Each market starts from a preset.
Educational only, not financial advice. Studies describe a chart; they don't predict it.