Summary
McGinleyDynamic is RTTA's streaming McGinley Dynamic: an adaptive moving
average that automatically speeds up when price runs away from the line and
slows when price is nearby, reducing whipsaw relative to a fixed-speed EMA.
Update API
result = rtta.McGinleyDynamic(window=14, fillna=True).update(price)
| Parameter | Default | Meaning |
|---|---|---|
window |
14 |
Speed constant \(N\) |
fillna |
True |
If False, NaN until window samples |
update(price) returns the current McGinley Dynamic value as a scalar.
Theory Of Operation
John R. McGinley designed the Dynamic to track price more closely than simple or exponential averages without constant period retuning. The step toward price is divided by \(N (price/MD)^{4}\):
- When price is far above MD, the ratio \(>1\) and the denominator grows, but the large numerator still pulls MD up; the fourth-power response is tuned so the line "catches up" smoothly rather than lagging like a slow SMA.
- When price oscillates near MD, effective smoothing is heavier, limiting noise.
The first observation seeds \(MD = price\). A zero MD is reseeded to price; a non-finite denominator also reseeds.
Recurrence
Let \(z_t\) be price and \(N=\max(\texttt{window},1)\). Seed:
For \(t > 1\), if \(MD_{t-1} = 0\) set \(MD_t = z_t\); else with \(r_t = z_t / MD_{t-1}\) (safe-divide defaulting to 1):
When fillna=False and fewer than \(N\) samples have been seen, return NaN;
otherwise return \(MD_t\).
Implementation Notes
- Implemented in
src/rtta/indicator.cppinclass McGinleyDynamic. - Uses
safe_divide(price, value_, 1.0)for the ratio. - Output is a scalar
double.