NFL Trends & SDQL: Winless Teams and Subsets Research
A winless NFL team can look like an easy team to oppose. But does its record tell us whether the point spread offers value? This NFL trends study uses the simple SDQL query wins=0 and KillerSports’ new Subsets Optimizer to explore that question.
The broad sample returned only 51.3% against the spread and a reported −2.1% ROI. One subset looked substantially different: winless teams facing an opponent on short rest went 86–46–6 ATS, covering 65.2% of decided bets with a reported +24.4% ROI. That makes it a worthwhile research candidate. Further testing is needed before treating it as a betting rule.
For the broader framework, visit our NFL trends research hub. If you are learning the query language, start with our NFL SDQL guide and examples.
What the KillerSports Subsets Optimizer does
According to KillerSports’ Subsets announcement, the tool evaluates about 70 logical subsets of a query, including favorite and underdog roles, home and away games, and rest situations. It makes comparisons that would otherwise require repeatedly editing the original query.
That is especially useful when a broad trend hides different betting situations. A winless favorite and a winless road underdog may share the same season record while facing very different market expectations. The optimizer helps identify which differences deserve closer investigation.
The base NFL SDQL query: wins=0
Select the NFL database in KillerSports, then enter:
wins=0The supplied query output showed:
| Measure | Historical output |
|---|---|
| Straight-up record | 1,174–1,433–9 (45.0%) |
| ATS record | 1,302–1,238–76 (51.3%) |
| Reported ATS profit | −$5,980 |
| Reported ATS ROI | −2.1% |
| Average spread | +1.5 |
| Average total | 44.5 |
These results do not support backing every winless team. The sample covered more often than it failed to cover, but that was insufficient to overcome the betting cost.
Source and scope: These figures reproduce a KillerSports output supplied for this study on October 4, 2026. The output did not identify its starting season, final included game date, or spread source. Profit and ROI are the tool’s reported figures, not independently audited returns or ProComputerGambler’s subscriber results. Records use wins–losses–pushes for ATS; cover percentages exclude pushes.
Winless-team NFL trends: the subset results
The following selection includes promising candidates and contrasting results. It is not a complete list of every subset tested.
| Subset label | ATS record | Cover rate | Reported ROI |
|---|---|---|---|
| Foe Short Rest | 86–46–6 | 65.2% | +24.4% |
| Off a Home Game | 356–287–14 | 55.4% | +5.7% |
| Off B2B ATS Losses | 238–192–9 | 55.3% | +5.7% |
| Extra Rest | 106–85–6 | 55.5% | +5.9% |
| Away | 660–565–38 | 53.9% | +2.9% |
| Dog | 813–700–44 | 53.7% | +2.6% |
| Home | 634–664–37 | 48.8% | −6.8% |
| Fav | 468–518–32 | 47.5% | −9.4% |
These are overlapping samples. A game can qualify as Away, Dog, and Foe Short Rest simultaneously. Their records cannot be added together, and the combined filter’s performance cannot be inferred from the separate rows.
Why opponent short rest deserves the first review
The Foe Short Rest subset produced 86 ATS wins and 46 losses, plus six pushes. Its reported profit was +$3,540. Under a $110-risk-to-win-$100 assumption, 86 × $100 − 46 × $110 = $3,540, which matches that profit figure.
One possible explanation is that a poor season record dominates perceptions while the opponent’s reduced preparation time receives less attention. A winless team can remain weak and still exceed spread expectations. That explanation is a hypothesis; these historical results do not demonstrate why the pattern occurred.
Before reproducing this trend, click the subset’s Run button and copy the generated SDQL. The precise rest threshold and any additional conditions were not included in the supplied output. “Foe Short Rest” should not automatically be translated into “Thursday game” or an assumed number of rest days.
Check whether “winless” includes season openers
A team entering its first game has zero wins, too. The query wins=0 can therefore mix 0–0 teams with teams that have already lost games. The optimizer’s Losing Teams row contained substantially fewer observations than the base sample, making this distinction worth checking directly.
As a first comparison, exclude Week 1:
wins=0 and week>1Then inspect each qualifying team’s prior games to confirm that it has actually played and remains winless. Compare 0–1 teams with teams that have two or more losses. Those are different situations, and the original table does not establish how each performed.
Our separate winless road teams after September study uses a different filter. Its published historical record should not be substituted for any of the subsets here.
NFL SDQL examples for further testing
Keep the original query and test one change at a time. These instructional queries have not been rerun to generate the results above:
wins=0 and week>1 and A
wins=0 and week>1 and D
wins=0 and week>1 and p:HA selects away teams, D selects underdogs, and p:H refers to a previous home game. Confirm the provider’s definitions and how previous-game references behave at season boundaries. Adding week>1 changes the population, so these examples do not inherit the displayed subset records.
For the short-rest candidate, preserve the optimizer’s exact generated query. Compare fixed periods such as seasons before 2010, 2010–2019, and 2020 onward, then examine individual seasons. These retrospective checks assess stability; because the subset was selected using the full history, they are not untouched out-of-sample validation.
How to interpret ROI and p-values
At −110 odds, the break-even win rate is 52.38%. A 51.3% cover rate loses money under that pricing assumption. Even a profitable historical percentage must be assessed against the odds available today.
Several displayed p-values match a one-sided binomial test against a 50% win rate. For Foe Short Rest, that produces approximately 0.000316. Testing the same 86–46 record against the −110 break-even rate instead produces approximately 0.00203, assuming independent decided bets. A p-value is not the probability that a system will remain profitable.
Searching many subsets also creates more opportunities to find an impressive result by chance. As an illustration, dividing a 0.05 threshold by 70 comparisons gives approximately 0.000714. The short-rest result clears that threshold against 50%, but not against −110 break-even. The actual testing family includes the choices made throughout the research, and the subsets overlap. This simplified check does not establish or disprove a betting edge. See the NIST explanation of the Bonferroni principle.
Using these NFL trends in a current matchup
The short-rest angle is the first candidate to investigate. Before making it an actionable rule, document the exact query, date range, odds assumptions, season results, and future performance after the rule is fixed.
For a current qualifier, check quarterback availability, injuries, the available spread, key numbers, and market movement. Our NFL Sharp Money Tracker and NFL Reverse Line Movement Tracker provide additional market context. Agreement with a historical angle is not independent proof of an edge.
To explore our subscription options and NFL research resources, review NFL Raw Numbers and ProComputerGambler membership plans.
Frequently asked questions
What is NFL SDQL?
NFL SDQL uses Sports Data Query Language to filter a football database by conditions such as season wins, location, spread, and previous results. It makes a research question explicit and repeatable when the database and field definitions are documented.
Are winless NFL teams profitable against the spread?
The supplied wins=0 sample returned 1,302–1,238–76 ATS and a reported −2.1% ROI. Some subsets performed better historically, but that does not establish that winless teams generally offer a profitable wager.
Which subset looked strongest in this test?
Foe Short Rest stood out at 86–46–6 ATS, 65.2% covers, and a reported +24.4% ROI. Its exact query and performance across seasons still need verification.
Does the Subsets Optimizer create automatic picks?
It identifies historical differences within a query. A promising subset still needs clear definitions, price-aware analysis, stability checks, and validation on data not used to select its rules.
