Fantasy Joes

Why I built it

I was interested in fantasy football rankings, but making my own in a spreadsheet was boring. I also wanted ordinary players to have a way to show how good their rankings were. The established expert lists put a lot of weight on having a media presence before anyone sees your calls.

The first idea was simple. Show me two players and make me pick one. I found that I didn’t always know which player I preferred until I had to make that choice. Do it enough times, and I have a ranking built from my own decisions. The game can then score those rankings against NFL results.

What I built

Fantasy Joes started with those two-player picks. Both the draft and weekly games later changed to show six players at a time. You choose the three you’d take first, in order. One round captures several preferences, so your rankings change faster than they did with individual pairs. The ranking algorithm starts from where players tend to be drafted or their projected weekly points, then gives more weight to surprising choices as you play.

The live weekly game asks which players you expect to score most that week. You can try it for free without signing up. Saving your rankings and seeing how they scored against NFL results requires an account. The draft game and its rankings remain on the site too.

I had tried the pair-pick idea before the 2025 season with a freelance developer. It took three to five seconds to load the next choice, which made a quick game feel slow, so I stopped. In 2026 I rebuilt it myself with Claude Code, including the illustrated player cards and a Discord version.

I also built player comparison pages drawn from the game’s picks. A matchup page shows how the crowd chose between two players. It also embeds the six-player game, where visitors can make their own picks. By September 2026, the site had more than a thousand of these pages. Search brought people to them; turning those visits into regular play was harder.

Around the game I built a player-art pipeline that could refresh portraits after team changes, a Discord game, partner attribution, and a growth dashboard tracking where visitors became players. A daily job published comparison pages from the draft picks, and the share cards used the same live data. These systems let me run more of the product alone, even when they didn’t solve retention.

I built a real-money contest system too, with wallets, seven contest formats, and eligibility controls. I ran an adversarial security review, fixed its findings, and tested a paid entry end to end in July. The contest work never turned into a paying business.

Why I stepped back

I planned to make money through paid contests based on ranking accuracy. The product got search traffic and some people played a lot, but an August 23 production check found no organic paid entries. The underfilled contests were cancelled in September, and cash entry is off.

The harder question was whether enough people would spend the time to make their rankings feel personal. The business also needed them to pay for contests repeatedly. I doubted one set of rankings gave people much reason to do that. I also wasn’t sure I had built for a specific player I understood well enough. I stopped pushing it as a business before the 2026 season.

The free weekly game is still live. I maintain what is already there while putting my time into Hometown Coloring Book. I explain the results and the decision in the Fantasy Joes wrap-up.