“After I finish reading an analysis, there are still plenty of things I have to do myself.”

During the XT Chinese-language AMA, The Next Step in AI Trading: Exploring Agent Applications Through NeoSoul (DNA), 0x31ad.sun described a gap he has encountered with AI trading tools: a finished analysis still leaves a lot of work before a trade is complete. Someone has to verify the reasoning, track changes in the market, and decide whether to place an order when the entry conditions are met. Once the result is in, there is another job: figuring out why it happened.

When those tasks are spread across several tools, the user has to connect them. That is the problem he pays closest attention to when evaluating products such as NeoSoul.

Connecting the steps around a trade

0x31ad.sun wants to handle those steps in a single application. He had downloaded and tried the NeoSoul client and said he had seen what appeared to be a trade review feature. In his view, assessing whether the tool can reliably handle trading requires a longer, verifiable operating record.

He was specific about what a review should show: the information available at the time, the reason for entering the trade, the risk conditions that were set, and the eventual result. Those details belong together. A return figure alone cannot explain the trade.

He also described two familiar reactions among traders: crediting their judgment when they make money and blaming the market when they lose. A profitable trade may simply have benefited from favorable conditions. A loss does not necessarily mean the strategy has no value. Without going back through the process, it is hard to tell these situations apart.

The same applies to an AI agent. Users need to see what it did under different market conditions and why. When the original conditions no longer held, did it adjust or stop? A useful review feature leaves that process open to inspection.

What happens when an order’s status is unclear?

0x31ad.sun sees AI trading tools moving from helping with analysis to helping with execution. That shift introduces problems an analysis report does not have to resolve.

He offered several examples. What happens if a data source returns incorrect information? If an order has been submitted but the platform does not give a clear order status, how does the system respond? After consecutive losses, does it keep running the same strategy or pause promptly?

He raised these questions as a way to assess the tools. The ability to analyze a market, generate a strategy, and place orders automatically does not mean every part of the process has been accounted for. In live trading especially, users need to see how the tool actually responds when something goes wrong.

He therefore wants operating records to show losses as well as gains, with fees and execution details available for review. Comparing those records across market conditions should gradually make the tool’s capabilities clearer. Adding support for more markets also calls for fresh scrutiny. Trading mechanisms and risks differ, so performance in one market cannot simply stand in for performance in another.

After consecutive losses, go back through the records

An audience member asked what several losing trades in a row might mean: unfavorable market conditions, a flawed strategy, or limitations in the AI itself?

0x31ad.sun suggested examining those possibilities separately. He would start with the market conditions when the trades took place. A strategy suited to a range-bound market may lose repeatedly when prices begin moving in one direction. Changes in volatility and liquidity can also affect results. The first question is whether the conditions the strategy depends on still hold.

Then comes the strategy itself. Were the entry conditions reasonable? Was the position size appropriate? How much of the return went to fees and slippage? If the same strategy performs poorly over time under similar market conditions, the market cannot take the blame indefinitely. Its logic needs another look.

The agent’s judgment and execution need a separate check. Did it miss important information, contradict its earlier assessment, or depart from the rules it was supposed to follow? Did it keep placing orders after the conditions were no longer met? Comparing those decisions with execution prices, risk exposure, and outcomes can help pinpoint the problem.

That explains his emphasis on records. The size of a loss is only the starting point for a review. When problems with the market, strategy, and execution are lumped together, users cannot easily tell what to change.

After several consecutive losses, 0x31ad.sun said he would pause the strategy or reduce the capital allocated to it, then review the records. He opposed letting the system keep trading in the hope that it would earn back the losses.

---

This article is based on the XT Chinese-language AMA, The Next Step in AI Trading: Exploring Agent Applications Through NeoSoul (DNA), and reflects 0x31ad.sun’s personal views and observations from using the product. It is not investment advice. Quotes have been translated from Chinese.

Listen to the full replay.

About XT Exchange

Founded in 2018, XT Exchange is a global digital asset trading platform serving more than 12 million registered users across over 200 countries and regions. Its services span spot, margin and futures trading, TradFi market exposure, a real-world asset marketplace, and everyday payments through XT Pay. Guided by Xplore Crypto Trade with Trust, XT Exchange enters its ninth year with Build the NeXT, reflecting its focus on trust, broader market access, and practical uses for digital assets.

Join the XT Exchange Community: X (Twitter) | Telegram | Facebook | Instagram | LinkedIn | Medium | YouTube

Disclaimer: XT Exchange reserves the right, at its sole discretion, to modify, amend, or cancel this announcement at any time for any reason without prior notice.