Vendor Abandonment

Vendor abandonment is the dominant hidden tail risk in EA purchase decisions, more consequential than the strategy class risks buyers typically focus on. The mechanism: every EA's edge depends on ongoing vendor maintenance — for ML-based products through retraining, for rule-based products through parameter updates as market conditions evolve, for all products through user support during deployment issues. When the vendor abandons the product, that maintenance stops. The EA continues to function technically but its edge decays as the market evolves away from the conditions the EA was designed for.

Patterns of vendor abandonment:

• **Silent abandonment**: vendor stops responding to support emails and Discord channels but keeps marketing the product. Buyers can't get help with deployment issues but the product remains for sale. Worst pattern because new buyers don't realise the vendor is gone • **Gradual fadeout**: vendor reduces response frequency over 6-12 months, eventually stopping. Pattern is often associated with vendors moving on to other projects or business interests • **Discontinuation without notice**: vendor pulls the product from marketplaces without warning to existing customers. Buyers lose access to updates and support; existing deployments continue but support disappears • **Brand pivot**: vendor rebrands under a new name with new products; previous brand's products are silently abandoned. Buyers must track vendor identity across rebrandings to understand which products are actually maintained

Why vendor abandonment is the dominant tail risk:

• **Higher frequency than blow-up**: surveys suggest 40-60% of commercial EAs are abandoned within 24 months of release. Strategy blow-up is more spectacular but less frequent • **Slow-onset failure**: unlike strategy blow-up (which is visible within weeks), abandonment failure plays out over months. The buyer often realises too late that the vendor is gone • **ML-amplified**: for ML-based EAs, abandonment essentially guarantees model decay because retraining stops. The EA continues to produce signals but the signals' edge decays month by month • **Difficult to reverse**: once a vendor has abandoned a product, restoration is rare. The buyer's options are to accept declining performance or replace the EA

Mitigations for vendor abandonment risk:

1. **Vendor longevity vetting**: check whether the vendor has been actively publishing for 18+ months across multiple products. New vendors carry higher abandonment risk regardless of product quality 2. **Update cadence verification**: confirm the vendor has shipped meaningful updates in the past 90 days. Vendors with longest gaps since last update are at highest abandonment risk 3. **Community presence**: vendors active in public forums, Discord, Telegram channels are less likely to be on the verge of abandonment than vendors with only website presence 4. **MyFXBook account freshness**: confirm the vendor's verified live account is updated within the past 30 days. Stale verification often precedes abandonment 5. **Multi-product portfolio**: vendors maintaining 2-3 products across different strategy classes have demonstrated business persistence that single-product vendors haven't 6. **Premium support tier availability**: vendors offering paid premium support tiers are economically incentivised to maintain responsiveness; vendors with no premium tier have weaker engagement incentives 7. **Realistic vendor commitments**: vendors who promise weekly model retraining without infrastructure capable of delivering it will eventually abandon; vendors who promise quarterly updates and consistently deliver them are operationally more sustainable

For EA buyer protection: budget for 12-18 month replacement cycles rather than indefinite deployment. Build the assumption of eventual vendor abandonment into the purchase decision; pay only what makes sense for 12-18 months of expected operation rather than for indefinite operation. Diversify across vendors rather than concentrating capital with single vendor.