Stage: Retired. Version v1; the hub record is as of 6 Nov 2026.
The verdict
NOT RELIABLE as a trend-following signal: the stack classifier was tested on ~1.2M symbol-days with no usable gradient; correctly retired. Episode-level held-rate odds remain a descriptive base rate, not independently re-tested here.
Development summary: The band-test episode/held-rate design is a reasonable EB-pooled base rate, but the stack-alignment classifier was measured on ~1.2M symbol-days to have no forward-return gradient — slightly inverse — and was retired as a signal.
What the review allows
MA-test episode flags (side, duration) and their EB-blended held-rate odds as descriptive trend-structure context.
The stack bucket as a partition/filter for browsing the universe by trend structure — not as a forecast.
What the review rules out
Trade or screen on the stack-alignment bucket as an entry edge — tested and retired; the small gradient present is inverse to the intuitive read.
Assume bull stacks continue or bear stacks keep falling — measured the opposite direction, weakly.
What it predicts
Item
Detail
Question it answers
Which names are currently testing an EMA 8/21/50 or SMA 100/200 as support or resistance, how long has the test run, and does the whole-stack alignment (bull/bear order) actually predict anything?
Method
MA band-test + EB pooling
How it works
Band-test episode (close within +-1 ATR of the MA after >=15 of prior 20 closes on one side) + held-rate EB pooling by MA x side, blended with the symbol's record; separate stack-alignment classifier
Inputs
EMA 8/21/50, SMA 100/200, ±1 ATR band test, 15 of 20 closes on one side, MA stack order (7 buckets), 8×21 EMA cross state
Headline results from the model's development record, as the hub publishes them.
Measure
Result
Compared with
Note
Stack-alignment forward-return gradient
none
an implied trend-continuation edge
~1.2M symbol-day backtest
Bear-stack vs bull-stack continuation
bear stacks bounced harder
bull stacks continuing
the small gradient present runs inverse to the naive expectation
8x21 EMA cross fold-in
cross direction + days since flip
the retired standalone EMA 8/21 tab
folded in as a column, not re-tested independently here
The hub's standard metrics for this model class (only those the record measured).
Metric
Sample / window
Compared with
Value
Range (CI)
Observations
symbol-days · undated
n/a
1,200,000
n/a
What the validation review found
Is the design sound? Reasonable split: an EB-pooled episode/held-rate descriptor is conceptually modest and appropriately framed; the stack classifier's ambition (predict from whole-stack order) was tested and correctly walked back.
Backtest. ~1.2M symbol-day test of the stack classifier is cited but undated in reviewed evidence; no separate backtest of the episode-level odds was found.
Limitations
The stack-alignment classifier has almost no forward-return gradient and what little exists runs slightly inverse (bear stacks bounce harder than bull stacks continue) — retired as a signal.
Episode-level Hold/Up/Exp odds were not independently re-validated in this review beyond noting they are EB base rates, not a fitted model.
The 1.2M-symbol-day stack backtest is undated in the evidence reviewed here.
Questions
Is the Moving-Average Support/Resistance Episodes validated?
Its validation review dated 3 Sep 2026 rated it NOT RELIABLE as a trend-following signal: the stack classifier was tested on ~1.2M symbol-days with no usable gradient; correctly retired. Episode-level held-rate odds remain a descriptive base rate, not independently re-tested here.
What does the Moving-Average Support/Resistance Episodes predict?
It answers: Which names are currently testing an EMA 8/21/50 or SMA 100/200 as support or resistance, how long has the test run, and does the whole-stack alignment (bull/bear order) actually predict anything? Output: Per-symbol MA-test episode (side, testing_d, new-flag) with EB-blended Hold/Up/Exp odds; stack-alignment bucket (descriptive only). Horizon: 10-session held/break outcome per episode. Coverage: ~1,700-name panel (S&P 500 + Nifty + curated watchlists).
How was the Moving-Average Support/Resistance Episodes tested?
Stack-alignment forward-return gradient (~1.2M symbol-day backtest): none; slightly inverse; Bear-stack vs bull-stack continuation: bear stacks bounced harder than bull stacks continued; Status: retired as a signal; descriptive episode odds still shown.