WinPicks is versatile enough to implement almost any handicapping method you can think of. Developing a method requires two things: deciding how many weeks of stats to use, and deciding which prediction formula to use.

Pro football handicappers often debate the number of weeks of stats to use. It's tempting to go back as far as possible (particularly to look for trends), but it isn’t always wise. Teams go up and down from season to season. Part of this is because the NFL gives last year’s division winners the most difficult schedules the following season. This makes it difficult for them to repeat as champions. Teams that finish last in their division are usually given an easier schedule, which often means they turn things around the following season. Another key factor is free agency. It’s common for a team to change from 30% to 50% of their players between seasons. Some say that this makes it more difficult to be a fan, and it obviously makes it more difficult to be a handicapper.

To accurately handicap the NFL, you need to use a sample of games that reflects the way a team is playing now. For this reason, WinPicks lets you use a rolling average that only looks at a team's most recent games. Most handicappers use a rolling average from the last 3 to 8 games, or else use the season-to-date totals.

Before choosing a prediction formula, we hope you try to create your own. We supply three formulas with WinPicks that have worked well in past seasons, but your own formulas might do better. The ANALYZE option lets you instantly test any formula to see how well it works. When you create a formula, you should put most of the emphasis on the major stats like points scored, points per 100 yards, total yards, winning percentage, and so on. Stats like field goal percentage and yards per punt are usually less important. If you use them at all, assign them a low weight.

If you use the SureLock Formula Wizard, it does the formula design work for you. It automatically builds a complete formula by selecting the stats to use, and then computing the optimum percentage weights to assign to each stat.