# 光世証券 (8617.T) — per-stock model accuracy

> Model predictability decomposition: measures how accurately the ml-factor-lgbm-9f-jp-growth model family predicts this stock's returns within the jp-listed universe. High hit_rate_long does NOT mean 'good stock to buy' -- it means 'the model historically predicted this stock well when selecting it LONG'. This is a property of model-stock fit, not intrinsic stock quality.

- **Name**: 光世証券
- **Sector**: 金融（除く銀行）
- **Universe**: jp-listed
- **Decomposes**: `ml-factor-lgbm-9f-jp-growth` (strategy verdict: marginal) — https://qbuntu.ai/strategies/ml-factor-lgbm-9f-jp-growth
- **Computed asof**: 2026-10-01T02:33:09
- **HTML**: https://qbuntu.ai/stocks/8617.T
- **JSON**: https://qbuntu.ai/v1/stocks/8617.T
- **License**: CC-BY-4.0

## Accuracy (median across grid)

Median LONG hit rate is 0.0801 — below a coin flip.

| metric | median | p25 | p75 | n |
|---|---:|---:|---:|---:|
| hit_rate_long | 0.0801 | 0.0153 | 0.2333 | 12 |
| hit_rate_short | 1 | 1 | 1 | 9 |
| avg_return_long_bps | -201.9 | -219.2 | 2.5243 | 12 |
| avg_return_short_bps | 561.1 | 539.9 | 561.1 | 9 |
| selection_rate_long | 0.0025 | 0.0012 | 0.0051 | 12 |
| selection_rate_short | 0.0003 | 0.0001 | 0.0008 | 12 |

Hit rate ≈ 0.5 means the model's picks on this stock were near coin-flip. Returns in bps/trade. Metrics with n = 0 are omitted; n < 5 is a thin sample.

## PnL contribution

- long: -5.1571
- short: 3.7208
- total: -1.4363
- appearances: LONG 270 / SHORT 77 of 89088

---

Research, not investment advice. These metrics describe model–stock fit, not the stock's merit.
Cite as: qbuntu (2026). 8617.T per-stock accuracy. https://qbuntu.ai/stocks/8617.T