Ryan Pannell

Ryan Pannell has one of the better startup war stories in the ETF business. In 2019, his team built an early trading system, rigorously tested it, and watched it crush returns for five straight months. "The performance was so outrageous that we weren't mentioning it to anyone because we just thought they wouldn't believe it," he recalls. "We were sitting there going, how are we going to articulate this without looking like we're gambling or we're lying?"

Then month six happened. The system couldn't identify a deep sector rotation, and "it drove like a bus full of my money off a cliff. In one month, it wiped out almost all of the profit of the previous five." The team went dark , "some of the guys, I like to say, went off to their bedrooms and wouldn't come out for a week. They weren't responding to emails."

From TJ's Crash to the ARC System

They gave that first system a name: TJ. "TJ basically got drunk and drove off a cliff," Ryan says. "It's funny now. It was not funny then." But the failure was foundational. The guardrails had been set too far from where they should have been , the system was optimized purely for gross returns, essentially operating as a black box. "When you're using a black box system, that's what you're going to get , a 7,000% 10-year return, but you'll have two or three periods where you'll be down 70 to 90%."

That experience birthed what Ryan calls the ARC , their AI Risk Containment system. It's effectively an AI off-ramp: a layered risk mitigation framework that prevents the kind of catastrophic drawdown that TJ produced. "Now we don't come anywhere close to that outcome," he says. The lesson: if you're a responsible manager, you cannot put client money on a system optimized purely for maximum return, no matter how good the 10-year backtest looks. "Unless you're just some billionaire that wants to mess around and has money to throw at these things, you can't run a system like that."

Building Trading Systems from Scratch

Ryan's background is deeply technical. He was part of the team that built one of the first 16-core computer systems with 96 gigs of RAM. "We built our own daughterboards, our own riser cards. It was running Windows 2000 Server 64-bit, and we had to build it in a filing cabinet because there were no towers appropriately sized to take it. Drives were in subsequent drawers." That hardware mentality carries into how they approach their trading systems , building from first principles rather than buying off-the-shelf solutions.

The current system powering their ETF products (including DIP) is dramatically more sophisticated than TJ ever was. The core philosophy shifted from optimizing for return to optimizing for risk-adjusted outcomes. Every strategy they run through the system must answer two questions clearly: what specific objective are you trying to achieve, and how does it behave when things go wrong?

The Evolution from Failure to Product

Ryan frames the TJ disaster as essential to where the firm is today. "I always say I'm so grateful for that because it wasn't catastrophic. We were way up and then this thing basically took us back to normal, which is better than if it wiped us out." The key insight was that the guardrails were "way too far away from where they should have been. We were really optimizing for profit, gross return."

The ARC system addressed this by adding layers of risk mitigation , essentially building an AI-powered circuit breaker that recognizes when the system is entering dangerous territory and takes corrective action before losses compound. It's the difference between a trading system that optimizes for return and one that optimizes for survivability.

Managing Multiple Funds with AI Infrastructure

What makes Ryan's operation unusual is the scale of the underlying AI infrastructure supporting multiple products , private funds plus a growing roster of ETFs. The system has evolved from that filing-cabinet computer to an enterprise-grade platform, but the mentality remains the same: build it yourself, understand every component, and never trust a black box.

The TJ story is funny in retrospect, but it contains a serious lesson that applies to every systematic strategy on the market. The difference between a backtest hero and a viable investment product isn't the return , it's the guardrails. Ryan learned that lesson with his own money before ever taking client assets. That's the kind of tuition that builds better systems.

What separates Ryan's approach from many AI-driven strategies is the transparency about failure. Most systematic managers only show the success stories , the backtests that worked, the drawdowns that were avoided. Ryan leads with the catastrophe because it demonstrates what he learned. The ARC system exists specifically because TJ failed, and the guardrails are calibrated from real losses, not hypothetical scenarios. For advisors evaluating AI-powered strategies, the question isn't whether the system can generate returns , most can in favorable conditions. The question is what happens when the system encounters something it wasn't trained for, and whether the risk framework is built from experience or just backtesting.

Key Takeaways

• Ryan Pannell has one of the better startup war stories in the ETF business.
• In 2019, his team built an early trading system, rigorously tested it, and watched it crush returns for five straight months.
• "When you're using a black box system, that's what you're going to get , a 7,000% 10-year return, but you'll have two or three periods where you'll be down 70 to 90%." That experience birthed what Ryan calls the ARC , their AI Risk Containment system.
• The lesson: if you're a responsible manager, you cannot put client money on a system optimized purely for maximum return, no matter how good the 10-year backtest looks.

Listen to the full conversation on Spotify, Apple Podcasts, or YouTube.