/decisions · 2026-05-18 · investing · smart-money-mirror-v1

Deploy smart-money-mirror v1 paper portfolio ($15k initial across 6 names, $5k reserve) — go / revise / defer / archive?

First-track paper deployment per founder greenlight 2026-05-18. Mirror top-N positions across 7 tracked managers (Pabrai excluded — 13F lookup failed). Paper only; no live capital at risk.

Mirror top-N positions across our 7 tracked managers using deterministic selection rules. Initial $15k: MSFT / AMZN / GOOGL / TSM at 0.5R each (persistent darlings, top-10 in 3+ managers for 2+ quarters) + SNDK / INTC at 0.5R each (2026 Q1 multi-manager entries). $5k cash reserve for mid-Aug 2026 Q2 13F window. NO price-based exits. Quarterly rebalance on 13F filings; only mechanical exit is 2+ managers exiting same name same quarter. Honest framing: we're not out-thinking Druckenmiller, we're copying him with a 45-90 day disclosure lag — and the lag is the cost of access.

Why deploy now (the operational fit)

The /investing:smart-money-watch skill is already built. The 2026-05-17 aggregate 2-year backfill is on disk. The selection rules are deterministic and reduce to ~6 trades on day one. This deploys NOW because the prerequisite infra is shipped — we don't need to wait for the backtest harness to mature for memory v1.1 to ship the smart-money mirror.

The basket is also intentionally simple. The whole point of mirror trading is that we're not adding alpha — we're following allocators who already do, with discipline. Adding complexity to "improve" the strategy would defeat its purpose.

Bull case vs revision risk (the explicit tension)

The bull-case argument for v1: 13F-arbitrage academic literature (Cohen/Polk/Silli 2010, Wermers 2000) shows equal-weighted top-N holdings of skilled active managers can beat passive benchmarks net of lag. Our 7-manager curation is genuinely positive-sum (Druckenmiller, Berkshire, Russo, Akre, Tepper, Tiger Global, Marathon — all public track records, all skill-based selection). The 6-name basket is concentrated enough to capture the signal but diversified enough that single-manager error costs <17%. The "no price exit" rule pairs with quarterly rebalance to ride the holding period long enough that the disclosure lag doesn't dominate. SNDK + INTC overlap with memory thesis = independent confirmation from two skill axes.

The revision risk: persistent-darling slice is 100% mega-cap tech (MSFT / AMZN / GOOGL / TSM). Concentrated factor exposure to AI-infra demand. If that single factor breaks, the strategy breaks. The 45-90 day disclosure lag could be material on fast rotations — by mid-Aug we'd see 2026 Q2 holdings, but managers may have already trimmed by then. Top-10-by-dollar is a coarse proxy (Berkshire's #1 AAPL has been heavily trimmed and we'd still see it #1). And SNDK + INTC are fresh 2026 Q1 entries — could reverse next quarter.

The picks: APPROVE locks in the 6-name $15k initial deploy with $5k reserve. REVISE walks back specific parameters (drop a name, change the persistent-darling sizing, narrow the basket). ARCHIVE concedes mirror trading is the wrong shape for our setup (e.g., founder prefers a thesis-first stack-only approach). DEFER waits for 2026 Q2 13F confirmation before deploying.

Selection rules (deterministic)

SliceDefinitionNamesSizing
Persistent darlingsTop-10 in 3+ managers AND held 2+ consecutive quartersMSFT, AMZN, GOOGL, TSM0.5R each = $10k
Recent multi-manager additions2+ managers add or significantly increase same quarter, within last 2 quartersSNDK (Druck + Tepper Q1 26), INTC (Druck + Tiger Q1 26)0.5R each = $5k
Cash reserveHeld back for next 13F window (2026 Q2 mid-Aug)$5k

Initial ticker basket

TickerRoleTranche 1Max v1Signal source
MSFTPersistent darling0.5R ($2,500)1RTop-10 in 3 managers (Tepper, Druckenmiller, Tiger)
AMZNPersistent darling0.5R ($2,500)1RTop-10 in 3 managers (Tepper, Druckenmiller, Tiger)
GOOGLPersistent darling0.5R ($2,500)1RTop-10 in 3 managers (Tepper, Berkshire, Tiger)
TSMPersistent darling0.5R ($2,500)1RTop-10 in 3 managers (Tepper, Druckenmiller, Tiger)
SNDKQ1 26 multi-manager entry0.5R ($2,500)1RNew positions by Druckenmiller + Tepper
INTCQ1 26 multi-manager entry0.5R ($2,500)1RNew positions by Druckenmiller + Tiger Global

Total: 3.0R = $15,000 initial deploy + $5,000 cash reserve = $20k strategy notional.

Rebalance rules (mechanical, quarterly)

TriggerAction
NEW 2+ manager same-quarter addition on a name not at v1 capAdd 0.5R (cap at 1R per name)
2+ managers EXIT same name same quarterClose that position entirely (only mechanical exit rule)
Persistent darling drops from 3+ managers to 2 in top-10Trim to 0.5R (or hold if already there)
Persistent darling drops to 1 managerClose position entirely
Existing position unchanged in signalHold; no action
Cash reserve allows add but no qualifying new signalHold cash; do not force-deploy
Founder kill-switch ("halt mirror" / "pause smart-money")Immediate close all

Deploy parameters

ParamValue
ModePaper only (Alpaca paper sandbox)
1R unit$5,000 (per parent README)
Initial deploy3.0R = $15,000 (6 names × 0.5R each)
Cash reserve$5,000 (1.0R, for 2026 Q2 13F window mid-Aug)
Strategy notional cap$20,000 (founder-greenlit)
Max per name in v11R ($5,000)
Per-trade stopNone (signal-stop, not price-stop)
Profit trimsNone — only multi-manager exits trigger sells
Rebalance cadenceQuarterly, ~mid-Feb / mid-May / mid-Aug / mid-Nov (13F filing windows)
Quarterly review ownerRay drafts post-13F-window; founder approves batch; Ray executes
Kill switchFounder channel — "halt mirror" / "pause smart-money" = close all
Audit log location~/rdco-vault/01-projects/investing/positions/smart-money-mirror-v1/

Lock in

Approve — deploy paper per v1 locked parameters

Ray fires initial 3R tranche (MSFT 0.5R + AMZN 0.5R + GOOGL 0.5R + TSM 0.5R + SNDK 0.5R + INTC 0.5R = $15k paper) into Alpaca. Logs to positions/smart-money-mirror-v1/. $5k cash reserve held for 2026 Q2 13F window (~mid-Aug). Use the field for any deploy notes.

Approve + send
Revise — walk back specific parameters

v1 is close but a parameter needs adjusting (e.g. drop INTC because overlap with memory v1.1, smaller persistent darling sizing, add a name from idea-generation candidates). Name what; Ray re-files v1.1.

Revise + send
Archive — strategy does not deploy

If mirror trading is the wrong shape for the setup (e.g., founder prefers thesis-stack-only or wants to wait for backtest infra). Ray files as status: archived-pre-deploy with reason. No paper trade.

Archive + send
Defer — wait for next-quarter signal confirmation

Hold for 2026 Q2 13F window (~mid-Aug 2026) to confirm SNDK + INTC positions held/added by tracked managers before deploying. Or wait for a named catalyst.

Defer + send