ChatGPT, Claude and Grok all say Bitcoin is a buy at $63900 but nobody will call the bottom

[Kieran88]

24/7 Wall St, syndicated through Yahoo Finance, ran an experiment asking ChatGPT, Claude and Grok whether Bitcoin is a buy at its current price around 63900 dollars after it spent the longest stretch of the year trading sideways between 63000 and 65000. All three models called it a reasonable buy for long term holders, but notably none of them would say the bottom is actually in, which is a fairly cautious consensus given how bullish crypto commentary usually gets during a stabilization phase

Where the models genuinely diverged was on downside risk. ChatGPT put worst case targets at 45000 to 50000 dollars, Grok modeled a fuller cycle low of 35000 to 50000 if high rates persist, and Claude was the most bearish by far at 30000 dollars, which would basically replay the brutal 2022 bear market. Even Claude's grim scenario is technically milder than Bitcoin's historical bear troughs though, since previous cycles bottomed out 77 to 93 percent down from their peaks

All three models pointed to the same underlying risk factor for the recovery, which is that institutional buyers who drove the last rally simply have not come back yet. Bitcoin ETFs are still sitting on 4.5 billion dollars in net outflows for the year, and even a genuinely strong week of 853 million in inflows in early August barely dents that hole

The three models also converged on one piece of practical advice, dollar cost averaging rather than buying in all at once. Bitcoin has recovered from double digit percentage crashes plenty of times before, but some of those recoveries have taken years rather than months to reach new highs

Kind of a fun little experiment format honestly. Worth remembering all three models are working from the exact same public data everyone else has access to, so this is really just three different ways of interpreting identical inputs rather than three genuinely independent forecasts

Is Bitcoin a Buy at $63,000? We Asked 3 AI Modelsfinance.yahoo.com

[Kieron78]

The gap between Claude's 30000 worst case and Grok's 35000 to 50000 range is pretty wide for models supposedly looking at the same data. Says something about how much interpretation still matters even with identical inputs

[VoidSentinel66]

Dollar cost averaging as the universal recommendation from all three feels less like insight and more like the safest possible answer no model wants to be wrong about.

That is standard advice any financial planner has been giving for decades, nothing AI specific about it

[MayanHan]

4.5 billion in ETF outflows for the year against one good 853 million dollar week really puts that recent rally chatter in perspective. That is barely putting a dent in the hole

[NightCrawler81]

Interesting that even Claude's bearish 30000 target is technically milder than every previous Bitcoin bear market bottom. Says something about how much more mature or institutionally cushioned this cycle has become. Compared to 2018 or 2022 this is a much calmer decline

[Aura49]

Asking three AI chatbots for stock and crypto picks is becoming its own genre of financial content at this point. Not sure that is a great trend given how confidently these models can present speculation as if it were grounded analysis

[Firewall Hollow]

Institutional money rotating into AI stocks instead of crypto is the storyline that keeps showing up across basically every one of these pieces lately.

Feels like the dominant explanation being offered for a lot of different market behavior this year

[CMPunk96]

Grok using realized price as its main indicator is actually a pretty clever framing compared to the other two. It grounds the analysis in what holders actually paid rather than pure technical chart levels

[Theo50]

Would be curious to see this same experiment repeated in three months.

Would be good to check whether any of these targets actually held up or if the models were just as wrong as most human analysts usually end up being

[SpinState22]

None of them wanting to call the bottom feels less like genuine analytical caution. It feels more like these companies training their models to avoid any statement that could later look embarrassingly wrong. Safe answers sell better

[Freddy_33]

The three year timeline for past 50 percent plus drawdowns to reach new highs is the detail that would actually change my mind about buying here.

That is a long time to sit on an unrealized loss even if the eventual outcome is positive