HERMES RESEARCH — Crypto Digest

🧠 SYNTHESIS — today's LLM digest (opportunities · avoid · actions)

# 🔬 CRYPTO RESEARCH DIGEST — 2026-08-12 ## Actionable Synthesis **No confirmed positive-EV process anywhere. The paper book as a whole is bleeding, and every chain, every source, and every trade class shows negative net expectancy after cost. The only "winners" in the entire run are micro-N processes (n≤13) with fat-tailed single prints — none promotable.** This is a DEFENSE / WAIT market for the machine, not an aggression market. **Where the least-bad blood is (relative edge, all still negative):** - **Sources:** `accum` is the champion signal (net_ev −7.3%, WR 23.6%, N=27,057, CONFIDENT) — the least-destructive high-volume line. `sniper` (−11.3%, N=6,322) next. Avoid the volume monsters: `trending_multi` (−24.0%, N=51,610) and `accum_engine` (−24.0%, N=8,320) just burn size at scale. - **Chains:** `sol` is least-bad (−14.9%, N=62,428) but only because it's largest. The real signal is **Robinhood = kill zone** (−46.9% net_ev, WR 4.1%, N=8,432) — already $300 min-buy gated (MANUAL-260810), correct call, keep it gated or pause. - **Trade classes:** `POSITION` (−7.2%) and `SWING` (−10.5%) hold up far better than `SCALP` (−23.0%) / `SNIPER_LAUNCH` (−24.6%). The fast scalping classes are the compressors of the bleed. - **TA:** Wins correlate with `CCI breakout` (+0.56% avg) and `volume_ratio high` (+0.61%). The EMA-stack markers are dead (Δ<0.3pp). BUT TA coverage is only 17.6% of resolved — too sparse to trust. **Recommendations (data-grounded):** 1. KEEP aggression at 0.3×; do NOT raise until TA score > −0.20 AND a chain flips net-positive. 2. Concentrate any live paper flow toward `accum` + `POSITION/SWING` lifetimes; shed `SCALP`-heavy exposure and cut `trending_multi` volume. 3. Robinhood: stay hard-gated (≥$300) or fully paused — −46.9% EV is the worst single cell in the book. 4. TA Coverage gap (17.6% vs 100% goal) is the #1 measurement hole. It biases the TA read AND the signal ranking. Fix ta_capture before trusting any TA-promotion. 5. 0 dormant awakenings (24+ cycles) + serial-funder collapse — the stealth-money pullback that the copy-timing edge depends on is not materializing. Do not add fresh/dormant copy exposure on empty tape. ## Avoid-list (net EV, CONFIDENT unless noted) | Cell | N | WR% | net_ev | |---|---|---|---| | Robinhood chain | 8,432 | 4.1% | −46.9% | | side_frontrun | 974 | 14.2% | −42.0% | | trending_multi | 51,610 | 22.0% | −24.0% | | accum_engine | 8,320 | 31.9% | −24.0% | | SCALP class | 54,724 | 22.7% | −23.0% | | wallet_buysize_edge | 1,485 | 17.0% | −26.8% |

⚙ STRATEGY OPERATIONS — paper trades opened by method · rules + cadence

method	live gate / rule	1h	4h	day	week	month	all
fomo	rules: detector code / feed_bible.json	4	11	84	576	1198	1226
oggy_kol	rules: detector code / feed_bible.json	0	0	48	146	774	947
inversion_alpha	rules: detector code / feed_bible.json	4	4	41	64	64	64
side_walker	rules: detector code / feed_bible.json	1	4	38	462	877	882
momentum_scalp	rules: detector code / feed_bible.json	2	8	34	372	1706	1839
michacalls	rules: detector code / feed_bible.json	0	1	12	50	82	82
smartmoney	rules: detector code / feed_bible.json	0	2	11	49	622	819
sa_accum	rules: detector code / feed_bible.json	0	0	10	35	83	83
cipher	rules: detector code / feed_bible.json	0	0	8	36	133	139
wallet_buysize	rules: detector code / feed_bible.json	0	0	7	37	601	643
trench	rules: detector code / feed_bible.json	0	0	7	19	31	83
cluster	min_wallets=3 · max_rug_ratio=0.4	0	0	6	105	930	952
accum_engine	rules: detector code / feed_bible.json	0	1	5	39	381	491
known_conv	rules: detector code / feed_bible.json	0	0	5	60	464	464
oggy_boost	configured; expand strategy_config.json	0	1	5	35	167	180
floor_revival	rules: detector code / feed_bible.json	0	0	5	18	72	85
whalemover	window_minutes=45 · min_buy_usd=300	1	1	3	49	415	468
volume_spike	rules: detector code / feed_bible.json	0	0	3	26	77	87
wallet_intel	configured; expand strategy_config.json	0	0	2	8	94	136
sajad_second_leg	rules: detector code / feed_bible.json	1	1	1	3	3	3
accum_sig_flow_div	rules: detector code / feed_bible.json	0	0	1	1	1	1
fomo_curated	rules: detector code / feed_bible.json	0	0	0	1	914	955
trending	PAPER ENTRY: fresh single tx ≥$1,500 AND MC $50,000–$2,000,000; raw discovery remains active · per-chain champion/challenger scanner gates	0	0	0	3	417	485
control	rules: detector code / feed_bible.json	0	0	0	13	191	294
sniper	rules: detector code / feed_bible.json	0	0	0	8	203	240
xcluster	rules: detector code / feed_bible.json	0	0	0	4	131	144
alpha_brief	rules: detector code / feed_bible.json	0	0	0	4	67	119
known_dist	rules: detector code / feed_bible.json	0	0	0	0	47	47
dexlions_kol	rules: detector code / feed_bible.json	0	0	0	3	42	46
kien_conv	rules: detector code / feed_bible.json	0	0	0	1	36	43
wc:SPRAYER	rules: detector code / feed_bible.json	0	0	0	0	5	32
wallet_copy	rules: detector code / feed_bible.json	0	0	0	0	0	27
wc:shadow_dowville	rules: detector code / feed_bible.json	0	0	0	0	17	17
wc:SWING	rules: detector code / feed_bible.json	0	0	0	0	5	9
dormant_wake	rules: detector code / feed_bible.json	0	0	0	1	4	4
accum_sig_bundled	rules: detector code / feed_bible.json	0	0	0	2	2	2

🏆 ALPHA LEADERBOARD — strategy × source (N≥100, by avg ret · EV=net expectancy)

strategy	n	win%	tp1%	sl%	avg ret	EV	top20%
accum:S1	6,524	23.0	28.7	54.4	-1.46%	-4.9%	46.5%
accum:S2	706	22.5	23.5	49.4	-4.07%	-5.8%	36.2%
accum:unknown	19,829	23.8	14.6	43.6	-6.35%	-8.1%	38.2%
sniper:S1	6,321	17.2	7.1	27.1	-6.42%	-11.3%	16.2%
kien_convergence:unknown	198	23.2	16.2	48.5	-9.46%	-10.9%	14.7%
control:unknown	1,151	12.7	12.7	34.1	-9.90%	-14.7%	24.3%
cluster:S1	9,669	14.9	18.2	60.8	-13.59%	-16.0%	24.5%
accum_engine:unknown	8,325	31.9	5.8	30.3	-21.50%	-24.0%	16.1%
trending_multi:unknown	51,667	22.0	4.7	29.9	-21.76%	-24.0%	7.9%
wallet_buysize_edge:unknown	1,485	17.0	3.2	34.6	-24.99%	-26.8%	14.1%
side_frontrun:unknown	974	14.3	3.2	51.0	-40.24%	-41.9%	7.4%

⛓ CHAIN RISK/REWARD

chain	n	win%	sl%	avg ret	EV	top20%
eth	4,538	31.4	13.0	-10.81%	-15.1%	6.1%
sol	62,450	22.4	40.8	-12.25%	-14.9%	24.1%
base	9,535	25.1	21.0	-15.40%	-17.1%	7.2%
bsc	22,146	24.3	34.1	-16.95%	-19.0%	18.2%
robinhood	8,446	4.1	49.1	-44.91%	-46.9%	3.4%

💰 BUY-PROFILE SWEET SPOTS — best band per metric

metric	best band	n	win%	avg ret	top20%
buys_ge_1k	3-5	1,396	37.2	+1.15%	22.1%
buys_ge_2_5k	0	3,353	27.3	+0.52%	26.8%
max_buy_pct_mc	0.5-1%	1,816	37.6	+0.48%	8.9%
buys_ge_500	5-9	1,800	33.9	+0.44%	18.5%
top3_pct_mc	<2%	2,468	36.3	+0.27%	5.7%
multi_buy_wallets	8+	24,214	36.5	+0.03%	12.5%

📐 TA CONDITIONS × WINS — valid at-signal-time capture (pipeline rebuilt 260716; N compounds daily)

condition	n	avg ret	top20%	edge?
volume_ratio high	805	+0.60%	10.2%	🟡 weak
cci breakout	877	+0.53%	11.8%	🟡 weak
vwap_pct above	1,447	+0.16%	11.1%	🟡 weak
volume_ratio med	554	+0.15%	10.4%	🟡 weak
ema21_above_50 True	1,208	+0.12%	10.4%	🟡 weak
ema9_above_21 True	1,368	+0.05%	10.4%	🟡 weak
vwap_pct below	1,236	+0.02%	11.0%	🟡 weak
ema9_above_21 False	1,163	+0.02%	9.2%	🟡 weak
ema21_above_50 False	1,206	-0.01%	9.0%	🟡 weak
cci neutral	1,251	-0.21%	9.1%	🟡 weak
volume_ratio low	1,324	-0.23%	11.8%	🟡 weak
cci oversold	409	-0.30%	8.2%	🟡 weak

📡 ALL SOURCES

source	n	win%	tp1%	sl%	avg ret	EV	top20%
known_convergence	2	0.0	0.0	0.0	+0.00%	-1.8%	0.0%
base_launch	91	16.5	3.3	11.0	-2.62%	-5.5%	7.0%
accum	27,059	23.6	18.2	46.4	-5.11%	-7.3%	40.1%
sniper	6,322	17.2	7.1	27.1	-6.41%	-11.3%	16.2%
kien_convergence	198	23.2	16.2	48.5	-9.46%	-10.9%	14.7%
control	1,151	12.7	12.7	34.1	-9.90%	-14.7%	24.3%
cluster	9,669	14.9	18.2	60.8	-13.59%	-16.0%	24.5%
sm_convergence	1	0.0	0.0	100.0	-15.00%	-16.8%	-15.0%
oggy_boost	1	0.0	0.0	100.0	-15.00%	-17.1%	-15.0%
accum_engine	8,325	31.9	5.8	30.3	-21.50%	-24.0%	16.1%
trending_multi	51,667	22.0	4.7	29.9	-21.76%	-24.0%	7.9%
wallet_buysize_edge	1,485	17.0	3.2	34.6	-24.99%	-26.8%	14.1%
trench	17	11.8	17.6	88.2	-34.12%	-35.9%	18.3%
side_frontrun	974	14.3	3.2	51.0	-40.24%	-41.9%	7.4%
wallet_intel	6	0.0	0.0	66.7	-41.25%	-42.7%	-7.3%
inversion_alpha	8	12.5	0.0	50.0	-47.74%	-49.1%	9.0%
whalemover	25	12.0	8.0	68.0	-62.30%	-66.1%	30.2%
momentum_scalp	40	7.5	12.5	87.5	-69.14%	-72.6%	10.3%
fresh_buy_permutation	3	0.0	0.0	100.0	-71.67%	-74.1%	-15.0%
michacalls	18	11.1	11.1	94.4	-79.17%	-81.7%	-6.2%
oggy_v3	53	3.8	3.8	96.2	-86.89%	-90.5%	-36.8%

🧭 KARPATHY LOG — latest reviews

2026-08-12T08:35 — Review completed — no param changes recommended

2026-08-11T20:35 — Review completed — no param changes recommended

2026-08-11T08:36 — Review completed — no param changes recommended

2026-08-10T20:36 — Review completed — no param changes recommended