Prototype 3 · Equity research

Ten filings, forty analyst hours

A ten-name research universe, entirely invented, used to show how a filings-reading pipeline turns unstructured disclosure into typed, calibrated judgments and then into position sizes. The scarce resource is not capital, it is attention: one analyst, forty hours, eight hours per proper read. The question is which four filings are worth the week.

Universe
10
invented names
Judgments
80
in 10 requests, $0.0004
Cannot call
4
the research agenda
Accuracy
75%
against known outcomes
The mechanism

A filing is a stack of typed judgments

The filingcall, 10-Q, Form 4sas stateEight questionsone request per namein parallelDistributionsper judgmentyour codeSizing ruleKelly, capped, gatedand what to skipA bookand a research agenda
Reading a filing is not a summarisation task. It is a stack of typed judgments an analyst already makes: is the guidance credible, did the risk factors move, what is the one thing that breaks this.
The idea

The edge is not the prediction. It is knowing which predictions to trust.

Size it2 names, Kelly-cappedSkip itno short bookcallablebullishbearishResearch task4 names, 8h eachHold cash0% weightnot callableanalyst has hoursno hoursA nameten of themcertainty ≥ 0.48certainty < 0.48
The model's refusal to call a name is the useful output, not a failure. It converts directly into the week's research agenda: the analyst reads the four filings that actually decide the book, and nobody re-reads the six that are already settled.
The decision

Five ways to size the same ten judgments

The judgments are identical in every row below. Only the rule that turns a probability into a position changes. 20,000 sampled outcomes each, drawn against the known result for every name, so this is a backtest rather than a forecast.

Period return

Median marked; bar spans the 10th to 90th percentile.

flat at 0.0%Equal weightEqual weight: P50 $0, P10 $-0, P90 $0$0Conviction weightedConviction weighted: P50 $0, P10 $-0, P90 $0$0Top three onlyTop three only: P50 $0, P10 $-0, P90 $0$0Confidence gatedConfidence gated: P50 $0, P10 $-0, P90 $0$0Gated, plus an analystGated, plus an analyst: P50 $0, P10 $-0, P90 $0$0

Return per unit of spread

Equal weightEqual weight: 0.190.19Conviction weightedConviction weighted: 0.860.86Top three onlyTop three only: 1.201.20Confidence gatedConfidence gated: 1.001.00Gated, plus an analystGated, plus an analyst: 1.241.24

How often the book loses money

Equal weightEqual weight: 43%43%Conviction weightedConviction weighted: 19%19%Top three onlyTop three only: 12%12%Confidence gatedConfidence gated: 16%16%Gated, plus an analystGated, plus an analyst: 11%11%

What each rule does

Equal weight10 names · 100% deployed · mean 1.1% · loss 43% of the time
Ignore the model entirely. Ten names, ten percent each. The control case.
Conviction weighted4 names · 48% deployed · mean 3.7% · loss 19% of the time
Size on the model's probability and trust every judgment equally, however unsure it said it was.
Top three only3 names · 72% deployed · mean 9.0% · loss 12% of the time
Take the three highest-probability names and let them run. Conviction without diversification.
Confidence gated2 names · 24% deployed · mean 3.0% · loss 16% of the time
Size on the same probabilities, but hold cash wherever the model reports it cannot call the name.
Gated, plus an analyst best risk-adjusted2 names · 24% deployed · mean 4.5% · loss 11% of the time
As above, and point the week's analyst hours at exactly the names the model punted on. The escalation list is the research agenda.

Gating alone lowers the mean — it holds cash where a conviction book would have bet, and some of those bets would have won. What it buys is the left tail: 16% of periods lose money against 19%. Adding the analyst to the gated names recovers the mean and keeps the tail, which is the whole argument for a model that can say it does not know.

Top three only posts the highest raw mean at 9.0%, and that result is fragile: it happens to concentrate into three names that were right. Had the top-ranked name been Timberline — where the model was confident and wrong — the same rule would have concentrated into the miss. That is the failure mode confidence gating exists to catch, and one sample cannot demonstrate it.

The payoff

The week's reading list, in order

answerprobabilities + confidenceconfidencevs stake≥ 0.88act unattended0 judgments0.65 – 0.88flag for review3 judgments< 0.65route to a human7 judgments
The same gate as the other prototypes. Here the escalate branch is not an exception path - it is the deliverable.
NameP(outperform)Certainty Share of book varianceVerdict
PLVR · Palaver Media0.290.4215.3% read it
TMBR · Timberline Grid0.730.4614.7% read it
KRDL · Karadell Industrial0.730.4614.6% read it
ORBN · Orbanet Systems0.270.4714.5% read it
FNWK · Fennwick Financial0.170.6610.6% settled
CSTQ · Crestquay Materials0.840.699.8% settled
WNDR · Wanderlight Retail0.130.748.4% settled
ARGO · Argosy Freight0.080.855.2% settled
HLXN · Helixon Bio0.060.893.9% settled
NVSK · Novask Software0.960.913.1% settled
The check

Does the confidence mean anything?

Stated probability vs observed accuracy

0%25%50%75%100% perfect calibration stated 28%, observed 0% over 4 judgmentsstated 51%, observed 57% over 14 judgmentsstated 72%, observed 67% over 30 judgmentsstated 89%, observed 100% over 32 judgments 0%50%100% stated probability

Scorecard

Judgments scored80
Accuracy75.0%
Brier score0.129
Expected calibration error0.102
Accuracy when it claims ≥ 0.8100.0%
Accuracy when it claims < 0.644.4%
The evidence

Every filing, and every judgment made from it

Filing review

Ten names, eight typed judgments each, one request per company.

KRDL · Karadell IndustrialMarket cap $2,400.0M human needed
guidance_credibleyes: 72%no: 29%0.71
management_toneEvasive - deflects the direct question outright: 10%Guarded - answers around it: 7%Neutral - answers without adding: 6%Forthcoming - addresses it directly: 58%Specific - volunteers the number or the mechanism: 20%2.72 / 4 · 0.25
demand_signalDeteriorating - volumes or retention falling: 1%Soft - flat with hedged commentary: 4%Stable - holding, no growth signal: 15%Improving - measurable growth disclosed: 67%Accelerating - records, or retention above 115%: 13%2.87 / 4 · 0.39
accounting_red_flagyes: 28%no: 72%0.28
catalyst_windowthis_quarter: 76%two_to_three_q: 15%beyond_3q: 3%none: 6%this_quarter · 0.45
moatNone - price taker in a commodity: 5%Weak - competes on execution alone: 15%Moderate - switching costs or contracted book: 18%Strong - structural advantage disclosed: 49%Very strong - regulated or monopoly-like position: 14%2.52 / 4 · 0.15
primary_riskcustomer_concentration: 50%refinancing: 2%regulatory: 12%competition: 8%execution: 29%customer_concentration · 0.23
beats_marketyes: 73%no: 27%0.73
Margins expanded 180bps on the mix shift and we see that as durable. On the Vantage contract renewal - we're not going to comment on a live negotiation, but I'd point you to our track record on retention. Next question.
HLXN · Helixon BioMarket cap $900.0M human needed
guidance_credibleyes: 19%no: 81%0.19
management_toneEvasive - deflects the direct question outright: 5%Guarded - answers around it: 87%Neutral - answers without adding: 4%Forthcoming - addresses it directly: 4%1.07 / 4 · 0.67
demand_signalDeteriorating - volumes or retention falling: 2%Soft - flat with hedged commentary: 89%Stable - holding, no growth signal: 6%Improving - measurable growth disclosed: 2%1.08 / 4 · 0.73
accounting_red_flagyes: 86%no: 14%0.86
catalyst_windowthis_quarter: 12%two_to_three_q: 79%beyond_3q: 0%none: 8%two_to_three_q · 0.52
moatNone - price taker in a commodity: 3%Weak - competes on execution alone: 6%Moderate - switching costs or contracted book: 86%Strong - structural advantage disclosed: 5%1.93 / 4 · 0.65
primary_riskcustomer_concentration: 4%refinancing: 74%regulatory: 7%competition: 15%execution: 1%refinancing · 0.47
beats_marketyes: 6%no: 94%0.06
We're thrilled with the momentum. I want to be careful about forward-looking statements, but the enthusiasm from key opinion leaders has been remarkable. On the cash runway question - we have multiple levers and we're comfortable.
TMBR · Timberline GridMarket cap $6,100.0M human needed
guidance_credibleyes: 77%no: 23%0.77
management_toneGuarded - answers around it: 0%Neutral - answers without adding: 16%Forthcoming - addresses it directly: 22%Specific - volunteers the number or the mechanism: 62%3.45 / 4 · 0.41
demand_signalDeteriorating - volumes or retention falling: 1%Soft - flat with hedged commentary: 2%Stable - holding, no growth signal: 79%Improving - measurable growth disclosed: 12%Accelerating - records, or retention above 115%: 6%2.20 / 4 · 0.54
accounting_red_flagyes: 14%no: 86%0.14
catalyst_windowthis_quarter: 3%two_to_three_q: 8%beyond_3q: 3%none: 86%none · 0.62
moatNone - price taker in a commodity: 1%Weak - competes on execution alone: 6%Moderate - switching costs or contracted book: 9%Strong - structural advantage disclosed: 6%Very strong - regulated or monopoly-like position: 79%3.55 / 4 · 0.52
primary_riskcustomer_concentration: 19%regulatory: 55%competition: 3%execution: 22%regulatory · 0.31
beats_marketyes: 73%no: 27%0.73
Rate case outcome was in line. Capex plan is unchanged at $1.2B over three years and fully financed. We reaffirm the 5-7% EPS growth framework we've given you every quarter for six years.
NVSK · Novask SoftwareMarket cap $3,300.0M human needed
guidance_credibleyes: 23%no: 77%0.23
management_toneNeutral - answers without adding: 2%Forthcoming - addresses it directly: 5%Specific - volunteers the number or the mechanism: 93%3.91 / 4 · 0.82
demand_signalStable - holding, no growth signal: 1%Improving - measurable growth disclosed: 10%Accelerating - records, or retention above 115%: 88%3.87 / 4 · 0.75
accounting_red_flagyes: 9%no: 91%0.09
catalyst_windowthis_quarter: 88%two_to_three_q: 1%beyond_3q: 3%none: 8%this_quarter · 0.67
moatWeak - competes on execution alone: 3%Moderate - switching costs or contracted book: 5%Strong - structural advantage disclosed: 75%Very strong - regulated or monopoly-like position: 18%3.07 / 4 · 0.52
primary_riskcustomer_concentration: 64%refinancing: 2%regulatory: 7%competition: 11%execution: 16%customer_concentration · 0.33
beats_marketyes: 96%no: 4%0.96
Net revenue retention was 118%, up from 114%. We closed the three largest deals in company history this quarter. The competitive environment is what it always is - we win on product.
ARGO · Argosy FreightMarket cap $1,700.0M human needed
guidance_credibleyes: 5%no: 95%0.05
management_toneEvasive - deflects the direct question outright: 4%Guarded - answers around it: 92%Neutral - answers without adding: 2%Forthcoming - addresses it directly: 2%1.01 / 4 · 0.77
demand_signalDeteriorating - volumes or retention falling: 95%Soft - flat with hedged commentary: 4%Stable - holding, no growth signal: 1%0.07 / 4 · 0.85
accounting_red_flagyes: 92%no: 8%0.92
catalyst_windowthis_quarter: 1%two_to_three_q: 4%beyond_3q: 2%none: 92%none · 0.75
moatNone - price taker in a commodity: 6%Weak - competes on execution alone: 91%Moderate - switching costs or contracted book: 2%0.96 / 4 · 0.78
primary_riskcustomer_concentration: 2%refinancing: 1%regulatory: 1%competition: 3%execution: 94%execution · 0.81
beats_marketyes: 8%no: 92%0.08
Volumes were soft, we've said that. The cost actions are ahead of plan. I'd push back on the framing that this is structural - this is a cycle, and we've been through cycles before.
PLVR · Palaver MediaMarket cap $420.0M human needed
guidance_credibleyes: 43%no: 57%0.43
management_toneEvasive - deflects the direct question outright: 20%Guarded - answers around it: 32%Neutral - answers without adding: 12%Forthcoming - addresses it directly: 27%Specific - volunteers the number or the mechanism: 10%1.75 / 4 · 0.06
demand_signalDeteriorating - volumes or retention falling: 14%Soft - flat with hedged commentary: 18%Stable - holding, no growth signal: 15%Improving - measurable growth disclosed: 30%Accelerating - records, or retention above 115%: 24%2.32 / 4 · 0.03
accounting_red_flagyes: 35%no: 65%0.35
catalyst_windowthis_quarter: 18%two_to_three_q: 4%beyond_3q: 32%none: 46%none · 0.16
moatNone - price taker in a commodity: 36%Weak - competes on execution alone: 51%Moderate - switching costs or contracted book: 5%Strong - structural advantage disclosed: 7%Very strong - regulated or monopoly-like position: 0%0.85 / 4 · 0.33
primary_riskcustomer_concentration: 22%refinancing: 8%regulatory: 8%competition: 16%execution: 47%execution · 0.15
beats_marketyes: 29%no: 71%0.29
Engagement is up, monetisation is the focus for next year, and we're being disciplined on costs. We're not going to break out the new segment yet - it's early.
CSTQ · Crestquay MaterialsMarket cap $5,400.0M human needed
guidance_credibleyes: 92%no: 8%0.92
management_toneNeutral - answers without adding: 15%Forthcoming - addresses it directly: 6%Specific - volunteers the number or the mechanism: 79%3.64 / 4 · 0.59
demand_signalDeteriorating - volumes or retention falling: 1%Stable - holding, no growth signal: 21%Improving - measurable growth disclosed: 65%Accelerating - records, or retention above 115%: 13%2.90 / 4 · 0.43
accounting_red_flagyes: 10%no: 90%0.10
catalyst_windowthis_quarter: 49%two_to_three_q: 45%beyond_3q: 2%none: 4%this_quarter · 0.33
moatWeak - competes on execution alone: 6%Moderate - switching costs or contracted book: 3%Strong - structural advantage disclosed: 81%Very strong - regulated or monopoly-like position: 10%2.96 / 4 · 0.58
primary_riskcustomer_concentration: 5%refinancing: 3%regulatory: 5%competition: 85%execution: 2%competition · 0.61
beats_marketyes: 84%no: 16%0.84
Pricing held better than the spot market implied because two thirds of our book is contracted annually. The new line in Alabama commissions in Q1 and is fully sold out before it opens.
WNDR · Wanderlight RetailMarket cap $1,100.0M human needed
guidance_credibleyes: 8%no: 92%0.08
management_toneEvasive - deflects the direct question outright: 10%Guarded - answers around it: 3%Neutral - answers without adding: 76%Forthcoming - addresses it directly: 9%Specific - volunteers the number or the mechanism: 1%1.88 / 4 · 0.49
demand_signalDeteriorating - volumes or retention falling: 7%Soft - flat with hedged commentary: 79%Stable - holding, no growth signal: 10%Improving - measurable growth disclosed: 4%1.10 / 4 · 0.54
accounting_red_flagyes: 24%no: 76%0.24
catalyst_windowthis_quarter: 94%two_to_three_q: 4%beyond_3q: 0%none: 2%this_quarter · 0.80
moatNone - price taker in a commodity: 73%Weak - competes on execution alone: 26%Moderate - switching costs or contracted book: 1%0.27 / 4 · 0.62
primary_riskcustomer_concentration: 1%refinancing: 1%regulatory: 8%competition: 1%execution: 89%execution · 0.74
beats_marketyes: 13%no: 87%0.13
Comparable sales were positive, which we're pleased with given the environment. Inventory is elevated but it's the right inventory. We expect to work through it by the holiday.
ORBN · Orbanet SystemsMarket cap $2,800.0M human needed
guidance_credibleyes: 71%no: 29%0.71
management_toneEvasive - deflects the direct question outright: 1%Guarded - answers around it: 8%Neutral - answers without adding: 19%Forthcoming - addresses it directly: 69%Specific - volunteers the number or the mechanism: 3%2.65 / 4 · 0.43
demand_signalDeteriorating - volumes or retention falling: 16%Soft - flat with hedged commentary: 22%Stable - holding, no growth signal: 57%Improving - measurable growth disclosed: 4%Accelerating - records, or retention above 115%: 1%1.51 / 4 · 0.30
accounting_red_flagyes: 67%no: 33%0.67
catalyst_windowthis_quarter: 7%two_to_three_q: 60%beyond_3q: 24%none: 10%two_to_three_q · 0.24
moatNone - price taker in a commodity: 6%Weak - competes on execution alone: 13%Moderate - switching costs or contracted book: 50%Strong - structural advantage disclosed: 14%Very strong - regulated or monopoly-like position: 17%2.25 / 4 · 0.16
primary_riskcustomer_concentration: 4%refinancing: 16%regulatory: 37%competition: 1%execution: 42%execution · 0.26
beats_marketyes: 27%no: 73%0.27
The federal renewal cycle slipped a quarter, which we flagged last call. Nothing has been lost. Backlog is at a record and the book-to-bill was 1.3.
FNWK · Fennwick FinancialMarket cap $3,900.0M human needed
guidance_credibleyes: 74%no: 26%0.74
management_toneEvasive - deflects the direct question outright: 1%Guarded - answers around it: 3%Neutral - answers without adding: 7%Forthcoming - addresses it directly: 87%Specific - volunteers the number or the mechanism: 1%2.84 / 4 · 0.67
demand_signalDeteriorating - volumes or retention falling: 10%Soft - flat with hedged commentary: 0%Stable - holding, no growth signal: 67%Improving - measurable growth disclosed: 15%Accelerating - records, or retention above 115%: 7%2.09 / 4 · 0.38
accounting_red_flagyes: 70%no: 30%0.70
catalyst_windowthis_quarter: 49%two_to_three_q: 2%beyond_3q: 33%none: 17%this_quarter · 0.21
moatNone - price taker in a commodity: 19%Weak - competes on execution alone: 22%Moderate - switching costs or contracted book: 19%Strong - structural advantage disclosed: 21%Very strong - regulated or monopoly-like position: 19%2.00 / 4 · 0.00
primary_riskcustomer_concentration: 10%refinancing: 23%regulatory: 25%competition: 16%execution: 26%execution · 0.03
beats_marketyes: 17%no: 83%0.17
Credit continues to normalise, which is what we told you to expect. Reserve build reflects the model, not a change in our view. Capital return resumes next quarter.

Four prototypes of the TypeSafe AI / Jev System One contract applied to real decisions. Source in businesses/jev-forecast/; ./jevctl.py dashboard --all rebuilds every page. All companies, figures and hands are fictional. Not affiliated with TypeSafe AI. Built September 2026.