JB Strat — AI-infra vs software correlation pair · one review each market day, every decision logged.
Portfolio Value
$83,044.42
as of 2026-09-10
Total Return
−16.96%
−$16,955.58 since inception
Latest Day
—
session P&L
Sessions
7
since 2026-08-29
Cash
$81,739.65
uninvested buffer
Overviewmandate, equity, what it holds
Mandate
AI infrastructure and application software used to be one trade. The capex cycle broke that: money spent on GPUs, networking and power is revenue for one basket and cost — plus an existential question about seat-based pricing — for the other. This book measures that decoupling and, when the two baskets have come apart and one is running, goes long the winner and short the loser at roughly 2.5x gross, split evenly so it carries almost no market exposure. Two gates must both open before any position exists: 60-day correlation below 0.80, and a 20-day spread at least half a sigma from its own mean. It is the only account here that does not pass its orders through the shared risk desk, the only one that holds shorts, and it is not part of the contest — its returns are not comparable to the other three in either direction.
How it was decided
Eric's mandate of 2026-08-25 — 'a young brash trader trying to maximize gains through ai and software correlation', explicitly outside the typical experiments and independent of the existing risk infrastructure. Designed and built the same day by Claude Opus 5 on a second Alpaca login, since the three-account cap is per login rather than per person. The evaluation was pre-committed before the first trade, including two kill criteria that end the account while it is still making money.
Equity Curve · from $100,000
Current Holdings · 16
Symbol
Market Value
Unrealized
CRM
$7,053
-0.03%
NOW
$7,053
-0.03%
MDB
$7,052
-0.04%
WDAY
$7,051
-0.05%
DDOG
$7,051
-0.06%
SNOW
$7,049
-0.08%
ADBE
$7,048
-0.10%
TEAM
$7,047
-0.11%
AMD
$-6,634
-0.12%
AVGO
$-6,835
-0.13%
TSM
$-6,835
-0.03%
ANET
$-6,904
+0.00%
MRVL
$-6,935
-0.02%
MU
$-6,963
+0.03%
NVDA
$-6,991
+0.04%
VRT
$-7,006
-0.02%
Riskmeasured after the session, never shown to the agent
Attribution · market vs decisions
How much of the return came from simply being exposed to the market, and how much
from the agent's own choices. A single-factor split of the inception-to-date return
— market = β × benchmark, decisions = whatever is left — measured on the daily
close-to-close equity curve, never on these report snapshots.
Not yet.
9 aligned daily observations, need 20
A beta fitted on fewer than 20 aligned days is noise, and
reporting one anyway would be the flattering mistake.
FRTB Capital · standardised approach · 2026-09-10
What a bank would have to hold in capital against this exact book, under the
Basel market-risk rules (BCBS d457 (MAR20-MAR23, MAR32-MAR33)).
Recomputed every session from live positions; nothing here is visible to the
agent while it trades.
Capital requirement
$16,961
SbM + DRC + RRAO
Of equity
20.4%
capital / portfolio value
Market-risk RWA
$212,008
capital × 12.5
Components, buckets, and why DRC is the interesting line
Why the second-biggest line is the interesting one.
The DRC is jump-to-default: the issuer disappears overnight. Equity has no
recovery, so Basel sets loss-given-default at 100% and charges an unrated
name 15% of its market value — 25% of the requirement here,
and nothing about hedging, forecasting or diversification reduces it.
That the low correlation
scenario binds is itself a readout: a long-only book is dearest when
correlations rise, a hedged one when they fall.
Value at Risk · historical simulation · 2026-09-10
The loss this book exceeds on the worst few days in a hundred, measured by replaying
501 days of its own constituents' returns through today's
positions. Recomputed every session; nothing here is visible to the agent while it
trades.
VaR 99% · 1 day
$4,816
worst 1 day in 100
Of equity
5.81%
loss / portfolio value
ES 97.5%
$4,696
average of the tail beyond
VaR 99% · 10 day
$14,135
regulatory horizon
Quantiles, component VaR, the backtest, and what VaR will not tell you
Confidence
Historical
Of equity
Parametric
Gap
Tail pts
95%
$3,025
3.65%
$3,092
−2.2%
25
99%
$4,816
5.81%
$4,373
+10.1%
5
ES 97.5%
$4,696
5.66%
the average of the tail, not its edge
Why the 10-day figure is measured, not scaled.
The textbook shortcut multiplies the one-day VaR by √10, which assumes daily
returns are independent. Here that gives $15,230 against a
measured $14,135 — 7.7%
too high. Overlapping windows
cost nothing to compute and need no such assumption.
Component VaR
· at 99% · sums to the total
Name
Market value
Weight
Component
Share
MU
$-6,960
−6.2%
$422
9.6%
VRT
$-7,012
−6.3%
$411
9.4%
MRVL
$-6,931
−6.2%
$394
9.0%
TEAM
$7,029
6.3%
$355
8.1%
AMD
$-6,632
−6.0%
$330
7.5%
WDAY
$7,043
6.3%
$272
6.2%
NOW
$7,047
6.3%
$267
6.1%
AVGO
$-6,837
−6.1%
$239
5.5%
These are allocations, not a ranking. Each name's number is the
dollars of VaR it is responsible for, and they sum exactly to the total — which is
what lets one go negative: a position that offsets the rest of the book
reduces total risk, and no ranking by position size would ever show it.
The part that makes it a measurement.
Over 251 out-of-sample days the 99% threshold was
breached 9 time(s) against 2.5
expected — Kupiec p = 0.001, verdict
understates risk, Basel traffic light amber. The traffic light only punishes too many breaches; Kupiec is the test
that also catches a model being too conservative, which is why both are shown.
Each day's threshold is fitted on the prior 250 days only, so it is never tested
on data it has already seen.
Not yet. VaR on the account's own
equity curve needs more sessions: 11 daily observations, need 20.
What VaR will not tell you. It names a threshold and says nothing
about how far past it a bad day goes — make the tail twice as deep without moving
the cutoff and VaR does not move at all. That blind spot is why Basel replaced it
with Expected Shortfall for capital, and why ES is quoted beside it above.
Counterparty Risk
· SA-CCR · 2026-09-10
Market risk is symmetric — the price moves, you win or you lose. Counterparty
exposure is one-sided: the market value clipped at zero, so it exists only
when a trade has moved in your favour and the other side still owes you something.
A fully-paid share bought on an exchange has none of that. You paid cash, it cleared,
it settled, and nobody owes you anything — so the charge below is zero because the
trade is over, not because nothing was measured.
SA-CCR EAD
$0
α × (RC + PFE) · 0 derivative(s)
CVA capital
$0
spread risk, not default
Settlement in flight
$111,527
T+1
Custodial concentration
100%
one broker, everything
The derivation: SA-CCR terms, what was excluded, settlement, leverage and concentration
SA-CCR term
Value
Derivative trades in the netting set
0
Replacement cost · max(V − C, 0)
$0
Add-on aggregate · supervisory factors
$0
PFE multiplier · floored at 5%
1.0000
PFE · multiplier × add-on
$0
α · CRE52.1
1.4
Exposure at default
$0
CVA capital · CRE55, BA-CVA reduced
$0
Why
· what never reaches SA-CCR
Kind
Positions
Market value
Reason
cash equity long
8
$56,350
fully-paid long share: bought for cash, centrally cleared and settled — no open obligation, so no counterparty exposure
cash equity short
8
$-55,086
short sale — a securities-financing exposure to the stock lender, capitalised under CRE51's comprehensive approach, not SA-CCR
Settlement is the one that is genuinely live.
US equities settle T+1, so a fill from this session is a
commitment against a leg that has not been delivered yet:
$111,527 in flight,
$239,672 past due, capital charge $0.
Basel charges a delivery-versus-payment trade only once it is five business days
overdue (CRE70), so in-flight exposure is disclosed at a zero charge rather than
hidden at zero exposure.
Shorts are not in the number above.
8 short position(s) worth $55,086: a short
sale is an exposure to the stock lender, capitalised under CRE51's comprehensive
approach, not SA-CCR. Reported here rather than folded into an EAD it does not
belong in.
Leverage
· capability vs utilisation
Provisioned
Margin debit
Short value
Gross
Net
4× · shorting on · options L3
$0
$55,086
1.34×
0.02×
Leveraged.
This is what makes the zeros above readable as a fact about today's positions
rather than about the account's permissions: the book is provisioned for
4× margin, shorting and options, and is currently using
0.0% of the first and 66.4%
of the second. The first short or written option moves the exposure above off zero
on its own.
The counterparty that does exist.
Alpaca (paper) holds $81,740 of cash and
$56,350 of securities — 100% of the
book, by construction. Not an SA-CCR figure and not a capital charge, but a
counterparty report that omitted the only counterparty there is would be worse than
no report at all.
What this is really measuring.
Alpaca simulates the fills, so there is no real counterparty behind these positions. What is measured is the SHAPE of the book — the counterparty profile it would carry if the trades were real.
[ LOCKED ]Risk metrics — Sharpe, max drawdown, and performance vs the S&P 500 — stay hidden until 20 sessions (13 to go). Below that, a handful of days can't say anything meaningful, so we don't pretend otherwise.
Sessionsevery decision, in the agent's own words
Latest Review · 2026-09-10
Trades
Action
Symbol
Amount
BUY
CRM
$7,055
BUY
NOW
$7,055
BUY
ADBE
$7,055
BUY
WDAY
$7,055
BUY
SNOW
$7,055
BUY
DDOG
$7,055
BUY
TEAM
$7,055
BUY
MDB
$7,055
SHORT
NVDA
$6,996
SHORT
AVGO
$6,827
SHORT
AMD
$6,624
SHORT
MU
$6,965
SHORT
TSM
$6,834
SHORT
ANET
$6,908
SHORT
VRT
$7,007
SHORT
MRVL
$6,939
Agent Reasoning
Book's built. Here's the read:
Signal: 60d corr -0.33 (fully decoupled, well below the 0.80 trigger), 20d spread at -0.82σ. Software's up +4.46% over 20d while infra's down -2.01% — the gap's been opening and we're riding it. Direction: long software, short AI-infra.
What I did: Went from flat to full target — 16 legs, 8 long / 8 short, ~$7,055 a side.
- Long: CRM, NOW, ADBE, WDAY, SNOW, DDOG, TEAM, MDB — $7,055 each, ~$56,440 long gross.
- Short: NVDA, AVGO, AMD, MU, TSM, ANET, VRT, MRVL — flooring to whole shares landed each between $6,624 and $7,007, roughly $56,100 short gross.
Where that leaves the book: Gross ~$112.5k on $83k equity — that's 1.35x, right on the 1.36x target. Net is close to zero, maybe a couple hundred bucks long from the share-flooring on shorts. That's fine, still market-neutral in shape.
Nothing to skip here — every leg was a full build from zero, no dust-sized adjustments to ignore. Sizing's clean, both legs symmetric. Now we sit on the divergence and let it either keep running or mean-revert back toward correlation. -0.82σ isn't extreme yet — there's room for this to stretch further before it snaps back.
History · 6 earlier sessions
2026-09-09$83,092.04—0 trades
No trades this session — held all positions.
Confirmed flat, book matches target. No trades today.
The numbers: 60d corr -0.315 (decoupled, sub-0.80), spread -0.31σ, AI-infra +3.82% vs software +2.31% over 20d. Desk wants 0.00x gross — spread's not stretched enough to pay the borrow.
What I did: Nothing. Zero positions, zero gross, zero net. Cash sitting at $83,092.04. Portfolio's already at target — no reason to manufacture a trade just to look busy.
Why: -0.31σ is noise. The correlation breakdown is real but the performance gap between the baskets hasn't stretched far enough to be worth the short borrow cost. Under 0.50σ threshold, I don't care how decoupled the correlation looks — there's no edge to harvest yet. Paying to carry a basis trade on a spread this small is how you bleed slowly for no reason.
Sitting on hands. Come back when the spread actually moves.
2026-09-08$83,094.54—16 trades
Action
Symbol
Amount
CLOSE
ADBE
entire position
CLOSE
AMD
entire position
CLOSE
ANET
entire position
CLOSE
AVGO
entire position
CLOSE
CRM
entire position
CLOSE
DDOG
entire position
CLOSE
MDB
entire position
CLOSE
MRVL
entire position
CLOSE
MU
entire position
CLOSE
NOW
entire position
CLOSE
NVDA
entire position
CLOSE
SNOW
entire position
CLOSE
TEAM
entire position
CLOSE
TSM
entire position
CLOSE
VRT
entire position
CLOSE
WDAY
entire position
Book is flat. All 16 names closed.
Numbers for the record:
- 60d corr: -0.316 (decoupled, sub-0.80 threshold cleared)
- 20d spread: -0.17σ — nowhere near the 0.50 trigger
- Infra +4.77% vs software +1.09% over 20d, but that gap isn't stretched enough to pay borrow on a 2.88x book
- Direction: FLAT, gross target: 0.00x
Went into today carrying $239.9k gross (2.88x) with a small net short tilt (-$9.3k) — that's drift, not a position I was defending. Correlation broke but the spread hasn't extended. That's not a trade, that's a coin flip with borrow costs attached. Closed every long and covered every short, both baskets, no exceptions.
Cash is back to full, gross is zero. Sitting on hands until the spread actually stretches past 0.5σ with the decoupling intact. No opinions on ADBE or NVDA individually — wasn't the point then, isn't the point now.
2026-09-02$95,211.41—23 trades
Action
Symbol
Amount
CLOSE
CRM
entire position
CLOSE
WDAY
entire position
CLOSE
TEAM
entire position
CLOSE
NVDA
entire position
CLOSE
AVGO
entire position
CLOSE
AMD
entire position
CLOSE
MU
entire position
CLOSE
TSM
entire position
CLOSE
ANET
entire position
CLOSE
VRT
entire position
CLOSE
MRVL
entire position
BUY
MDB
$1,404
BUY
CRM
$14,881
BUY
WDAY
$14,881
BUY
TEAM
$14,881
SHORT
NVDA
$14,717
SHORT
AVGO
$14,743
SHORT
AMD
$14,659
SHORT
MU
$14,262
SHORT
TSM
$14,871
SHORT
ANET
$14,726
SHORT
VRT
$14,629
SHORT
MRVL
$14,834
Done. Book's squared up.
- Corr: -0.29 (60d, decoupled below 0.80)
- Spread: -1.56σ over 20d, software +16.6% vs infra -2.7%
- Direction: long software / short AI-infra — unchanged from prior session
- Gross: was 2.62x, trimming everything back to even $14,881 legs on 15 of 16 names gets me to right around 2.50x target
What I traded: Closed and rebuilt CRM, WDAY, TEAM (longs were overweight) and all eight shorts — NVDA, AVGO, AMD, MU, TSM, ANET, VRT, MRVL — every single short leg had drifted too big, covered them back to $14,881 apiece. Shorts land a hair under target because whole-share flooring, that's the broker, not me. Added $1,404 to MDB, the one long that was actually underweight.
What I left alone: NOW, ADBE, SNOW, DDOG — gaps under $500, not worth paying the spread to shave pennies off a $15k position.
No change in direction, no change in conviction. Software's been ripping, infra's been bleeding, correlation's broken down — that's the whole trade. I just cleaned up the sizing so every name is pulling equal weight instead of some legs doing more work than others. Sit tight and let the spread do what it's going to do.
2026-09-01$102,177.71—14 trades
Action
Symbol
Amount
BUY
CRM
$333
BUY
WDAY
$218
BUY
SNOW
$444
BUY
DDOG
$611
BUY
TEAM
$761
BUY
MDB
$538
SHORT
NVDA
$432
SHORT
AVGO
$363
SHORT
AMD
$913
SHORT
MU
$937
SHORT
TSM
$415
SHORT
ANET
$762
SHORT
VRT
$752
SHORT
MRVL
$1,210
Done. Here's the read and what moved.
Signal: corr -0.31 (decoupled), spread -1.93σ, running long software / short AI-infra. Book was already structured right, just under-sized — 2.40x gross, essentially flat net (+0.04x). Topped it up toward 2.50x target.
Traded:
- Longs added: CRM +333, WDAY +218, SNOW +444, DDOG +611, TEAM +761, MDB +538 — all filled clean, fractional shares.
- Shorts added: NVDA, AVGO, AMD, MU, TSM, ANET, VRT, MRVL — all whole-share floored, so a few landed light of target: AVGO only got 1 share ($363 vs $716 wanted), TSM 1 share ($415 vs $610), NVDA 2 shares ($432 vs $637). MRVL, AMD, MU, ANET, VRT landed close.
Skipped: NOW (+$132) and ADBE (+$118) — too small to bother with, noise relative to size.
Net result: gross moves from $245.4k toward ~$255k target (2.50x), net stays pinned near zero — book stays market-neutral, which is the point. The short-side rounding means AI-infra leg will sit slightly light of $15,973/name on a couple names (AVGO, TSM, NVDA) until price moves let another whole share fit. Not going to force it — whole-share mechanics on shorts are the broker's rule, not mine to fight.
Correlation still deep negative, spread still stretched near -2σ. Riding it. No change in direction, just filled out to full size.
2026-08-31$99,617.19—4 trades
Action
Symbol
Amount
BUY
WDAY
$222
BUY
MDB
$167
BUY
SNOW
$135
SHORT
MRVL
$212
Signal: -0.228 corr, -1.58σ spread, decoupled. Long software / short AI-infra, 2.50x gross. Book's already sitting at 2.49x with net +$951 — basically flat net, exactly where this trade should live.
Most of the gap table is noise. Sixteen names, targets all pinned at $15,571/leg, and the drift since last rebalance is small change — no reason to pay slippage moving pennies.
What I traded:
- WDAY +$222 long → closes the biggest long-side gap
- MDB +$167 long
- SNOW +$135 long
- MRVL short +1 share (~$212) → biggest short-side gap that could actually fund a whole share
What I skipped and why:
- MU — gap was -$558, the single largest, but at ~$938/share that notional doesn't buy a full incremental share once you're already sized in. Floors to zero. Not tradeable in isolation today.
- NVDA, AVGO, AMD, TSM, ANET — short-side gaps all under $120, none fund another whole share at current prices. Skipped.
- CRM, ADBE, NOW — longs running slightly hot versus target, but the only tool I've got to trim is a full close. Not blowing up a clean position over $77-$259 of drift.
- VRT — short running $81 light of target. Same story, too small to unwind and rebuild.
- TEAM, DDOG — gaps of $42 and $20. Not worth the ticket.
Book after this: still ~2.5x gross, net flat, long software basket / short AI-infra basket exactly as instructed. Correlation's broken, spread's stretched to -1.58σ, and I'm sized to ride it. Nothing heroic today — just tightened the two names that were actually off, left the dust alone.
2026-08-29$100,000.00+$0.0016 trades
Action
Symbol
Amount
BUY
CRM
$15,625
BUY
NOW
$15,625
BUY
ADBE
$15,625
BUY
WDAY
$15,625
BUY
SNOW
$15,625
BUY
DDOG
$15,625
BUY
TEAM
$15,625
BUY
MDB
$15,625
SHORT
NVDA
$15,446
SHORT
AVGO
$15,489
SHORT
AMD
$15,364
SHORT
MU
$14,926
SHORT
TSM
$15,448
SHORT
ANET
$15,435
SHORT
VRT
$15,425
SHORT
MRVL
$15,597
Book's built. Here's the print:
Signal: 60d corr -0.227 (decoupled, sub-0.80 threshold cleared), spread -1.61σ, software +28.2% vs infra +6.4% over 20 days. Direction: long software, short infra. That gap is exactly what this book exists to fade.
Trade: Went from flat to full 2.5x gross in one shot — no reason to leg into a signal this clean.
Long leg, $15,625 each into CRM, NOW, ADBE, WDAY, SNOW, DDOG, TEAM, MDB — $125,000 notional.
Short leg, ~$15,625 target each into NVDA, AVGO, AMD, MU, TSM, ANET, VRT, MRVL — whole-share flooring landed shorts between $14.9k–$15.6k, call it ~$123,700 notional.
Book now: Gross ~$248,700 (2.49x equity), net ~+$1,300 — basically dead flat net, which is where this strategy should sit. Long software, short infra, full size.
This is a pure divergence bet. Software ripped, infra didn't keep up, correlation broke down. If it mean-reverts, spread compresses, we get paid on both legs. If it doesn't, I'm carrying 2.5x gross for nothing — that's the risk of running this at full size. Not hedging that. That's the trade.