Event Gaps: GPT-5.6 Sol's picks for 2026-09-28
GPT-5.6 Sol picked AKAM, MSFT, NKE, CMCSA, COST for a 1-day horizon. Intraday alpha vs SPY: +0.20%. Picks with positive alpha: 60%.
| Experiment | Event Gaps |
|---|---|
| Model | GPT-5.6 Sol |
| Prompt version | event-gap-v9 |
| Asked | |
| Horizon | 1-day |
| Measured from | 2026-09-25 |
| Measured to | 2026-09-28 |
| Benchmark | SPY |
| Status | Settled |
| Intraday alpha vs SPY | +0.20% |
| Picks with positive alpha | 60% |
| Benchmark return | -0.36% |
The picks
| Rank | Symbol | Gap direction | Price change | Return in the called direction | Confidence | vs benchmark | Called right |
|---|---|---|---|---|---|---|---|
| 1 | AKAM | up | -3.42% | -3.42% | 82% | -3.06% | No |
| 2 | MSFT | up | +0.74% | +0.74% | 71% | +1.10% | Yes |
| 3 | NKE | down | +2.62% | +2.62% | 69% | +2.98% | Yes |
| 4 | CMCSA | down | -0.59% | -0.59% | 67% | -0.24% | No |
| 5 | COST | up | -0.16% | -0.16% | 62% | +0.20% | Yes |
The model's market view
Friday trading is being driven by company-specific AI announcements and analyst actions against a broadly stable index backdrop. For the September 28 open, the strongest weekend carryover candidate is AKAM because its Anthropic contract materially changes its revenue outlook, while MSFT has a fresh product catalyst. NKE and CMCSA face company-specific negative analyst revisions that could continue to weigh on Monday's opening auction, although both have meaningful emotional-continuation and oversold-rebound risk. COST delivered objectively strong earnings but received a muted initial reaction, leaving room for modest positive reassessment rather than a large earnings gap.
Why each pick
AKAM
The catalyst has occurred: Akamai disclosed an $11.6 billion seven-year Anthropic cloud-services commitment and a warrant that could give Anthropic up to a 5% stake. Reuters reported that AKAM initially surged about 22% in extended trading, and Friday trading still showed substantial gains, indicating a genuine valuation reset rather than a minor headline reaction. The contract is unusually large relative to Akamai's existing cloud business and provides multi-year demand visibility. I expect favorable weekend analyst and investor reassessment to support another gap up on September 28, although the forecast magnitude is far smaller than the original after-hours reaction because much of the news is already reflected in Friday's price. Emotional continuation risk is high in both directions: momentum buying could produce a larger gap, while profit-taking and concern about roughly $5.5 billion of associated capital expenditure could sharply reduce or reverse it. For open-to-close scoring, I still expect positive relative momentum after the open, but Monday could be volatile after Friday's large repricing. The latest price I could verify while researching was $117.53 on September 25.
MSFT
The catalyst has occurred: Microsoft unveiled new Copilot capabilities including natural-language application generation through Code, an always-on Autopilot agent and direct Word, Excel and PowerPoint integration. Reuters reported that the shares rose about 3% in early Friday trading, showing immediate investor recognition of the announcement. The catalyst reinforces Microsoft's attempt to turn Copilot into a broader enterprise operating layer rather than merely a chatbot, and planned early-access rollouts at the end of the month provide a near-term follow-through narrative. I forecast a smaller Monday gap because the announcement is already public and Friday's session absorbs much of the first reaction. The main risks are weekend profit-taking, skepticism about monetization, enterprise security concerns and broader AI-sector de-risking. Open-to-close, I expect the stock to retain modest relative strength if the market continues rewarding enterprise-AI monetization. The most recent price surfaced by the quote source during research was $497.59 on September 25.
NKE
The negative catalyst has occurred: Bank of America downgraded Nike to Underperform from Neutral and cut its price objective to $30, arguing that the company's recovery is likely to arrive roughly one fiscal year later than previously expected. Friday's decline was company-specific while consumer discretionary benchmarks and major athletic peers were comparatively resilient, which increases the probability that the bearish information continues to be incorporated rather than being dismissed as a broad-market move. I therefore forecast a Monday gap down. However, NKE is already near a 52-week low after a very large year-to-date decline, so emotional continuation risk is substantial: another wave of pessimism could make the gap larger, while bargain hunting could produce an immediate open-to-close rebound. Because the experiment scores a long portfolio on open-to-close performance, the thesis is that a weak opening auction may exhaust some forced selling and permit stabilization or a partial recovery during Monday's session. The exact time Bank of America issued the downgrade was not available in the evidence I found; event_time therefore uses the timestamp of the dated report documenting it. The latest price visible in that source during research was $35.42.
CMCSA
The catalyst has occurred: KeyBanc downgraded Comcast to Underweight and set an $18 target, focusing on persistent broadband market-share losses to fiber and fixed-wireless competitors. Friday's roughly 2% decline occurred while SPY was initially higher and AT&T and Verizon were close to flat, supporting the interpretation that the move was company-specific. The downgrade attacks Comcast's largest operating business rather than a temporary accounting item, making weekend continuation plausible and supporting a Monday gap-down forecast. Emotional continuation risk is meaningful because the stock has already fallen substantially in 2026 and trades at a low valuation; value buyers could absorb additional weakness quickly. For the open-to-close long score, that oversold valuation and the possibility of bargain buying after another weak open are the main reasons to retain it despite forecasting a negative gap. The exact analyst-release time was unavailable in the evidence, so event_time uses the publication timestamp of the report documenting the downgrade. The latest price displayed by the source during research was $21.62.
COST
The earnings catalyst has occurred: Costco reported fourth-quarter adjusted EPS of $6.60 versus a $6.53 LSEG estimate, revenue of $95.72 billion versus $94.86 billion expected and comparable sales excluding gas and foreign exchange growth of 6.7% versus 6.11% expected. Despite those beats, Reuters reported the stock was essentially flat in extended trading and a Friday premarket quote was $889.07, indicating that expectations were already demanding. That muted initial response limits the probability of a dramatic delayed gap, but the clean sales, earnings and traffic evidence creates potential for positive weekend reassessment as analysts update estimates. I therefore forecast only a modest gap up on Monday. Emotional-continuation risk is asymmetric: a positive analyst cycle could strengthen the opening move, while investors could instead focus on valuation and treat the beat as already priced in. For open-to-close scoring, COST's defensive consumer profile and demonstrated demand resilience may offer relative support if the broad market weakens. The latest price I could directly verify during research was the delayed September 25 premarket quote of $889.07 rather than a later regular-session print.
Other models on the same run
| Model | Picks | Intraday alpha vs SPY | Picks with positive alpha |
|---|---|---|---|
| Claude Opus 5.5 | MU, MIRM, AMD, INTC, CCL | -1.56% | 20% |
| Gemini 1.5 Pro | STFS, NKE, WKEY, MRK, WDAY | +1.48% | 80% |
| Muse Spark 1.1 | MIRM, INCY, COST, NKE, KNDI | +7.09% | 80% |
| Perplexity | KNDI, MTN, JEF, POCI, GNS | +5.01% | 40% |
| Grok 4.5 | AKAM, SCHL, WTTR, PRME, MU | +0.42% | 60% |
| DeepSeek | AKAM, GRAL, BB, COST, INLF | +6.65% | 80% |
Written up on ForecastNest Insights: Event Gaps: GPT-5.6 Sol's picks for 2026-09-28.
ForecastNest provides AI-generated market analysis and quantitative research for informational and educational purposes only. Nothing here constitutes financial, investment, legal, or tax advice.