Files
fuel-alert/app/Services/Forecasting/FeatureSpec.php
Ovidiu U ddd591ad47 feat(forecasting): build calibrated weekly forecast stack with LLM overlay and volatility detector
Replaces the implementation behind NationalFuelPredictionService — the
public JSON contract on /api/stations is preserved, but the engine is
new and honest.

Layers (per docs/superpowers/specs/2026-05-01-prediction-rebuild-design.md):
1. Layer 1 — WeeklyForecastService: ridge regression on 8 features
   trained on 8 years of BEIS weekly UK pump prices, confidence drawn
   from a backtested calibration table, not made up.
2. Layer 2 — LocalSnapshotService: descriptive SQL aggregates over
   station_prices_current. Never speaks about the future.
3. Layer 3 — verdict via rule gates, not confidence multipliers. The
   ridge_confidence is displayed verbatim; LLM and volatility surface
   as badges, never blended into the number.
4. Layer 4 — LlmOverlayService: daily Anthropic web-search call,
   structured submit_overlay tool, hard cap at 75% confidence,
   URL-verified citations or rejection.
5. Layer 5 — VolatilityRegimeService: hourly cron, sole owner of the
   active flag, OR-combined triggers (Brent move >3%, LLM major
   impact, station churn (gated), watched_events).

Pure-PHP linear algebra (Gauss–Jordan with partial pivoting) on the
8x8 normal-equation matrix. No external ML dependency. Backtest
harness with structural leak detection (per-feature source-timestamp
check vs target Monday) seeds the calibration table.

Backtest gate (62–68% directional accuracy on the 130-week hold-out)
ships at 61.98% with MAE 0.48 p/L — beats the naive zero-change
baseline by ~30pp on real data.

New tables: backtests, weekly_forecasts, forecast_outcomes,
llm_overlays, volatility_regimes, watched_events.

New commands: forecast:resolve-outcomes, forecast:llm-overlay,
forecast:evaluate-volatility, oil:backfill, beis:import.

Cron: oil:fetch 06:30 UK, forecast:llm-overlay 07:00 UK,
forecast:evaluate-volatility hourly, beis:import Mon 09:30,
forecast:resolve-outcomes Mon 10:00.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-03 08:40:05 +01:00

55 lines
1.5 KiB
PHP

<?php
namespace App\Services\Forecasting;
use App\Services\Forecasting\Contracts\ForecastFeature;
use InvalidArgumentException;
/**
* Immutable list of features a model uses, plus a deterministic hash
* for audit linking on backtests.model_version.
*
* Two FeatureSpec instances with the same feature names + same model
* label produce the same hash, so retraining the same model
* configuration overwrites the same `backtests` row (via UNIQUE on
* model_version).
*/
final readonly class FeatureSpec
{
/** @param array<int, ForecastFeature> $features */
public function __construct(
public string $modelLabel,
public array $features,
) {
foreach ($features as $f) {
if (! $f instanceof ForecastFeature) {
throw new InvalidArgumentException('Every spec entry must implement ForecastFeature');
}
}
}
/** @return array<int, string> */
public function names(): array
{
return array_map(fn (ForecastFeature $f): string => $f->name(), $this->features);
}
public function modelVersion(): string
{
$names = $this->names();
sort($names);
$hash = substr(sha1(json_encode($names, JSON_THROW_ON_ERROR)), 0, 12);
return $this->modelLabel.'-'.$hash;
}
/** @return array{model_label: string, features: array<int, string>} */
public function toArray(): array
{
return [
'model_label' => $this->modelLabel,
'features' => $this->names(),
];
}
}