I am using 19.07 on TP-LINK TL-MR3220 v2 (don't judge me) with one broadband pppoe WAN (~ 70 Mbps, ~3 ms ping) and a smartphone-tethered unlimited 4G WAN (~30 Mbps, ~70 ms ping) as a fallback via MWAN3. Its current default "balanced" policy config implies all connections go through both interfaces in the configured weight proportion regardless their current bandwidth usage (i.e. with 60/40 config 6 of 10 connections go through the primary interface, and 4 of 10 go through the fallback one). The problem is a random service may need a fast connection but may be "unlucky" to be assigned to the slower mobile WAN interface - this is totally unpredictable, so I had to switch to "wan1_wan2" policy. With "wan1_wan2" policy I have connections going through wan2 only if wan1 is down, so wan2 is idling 99% of time.
The question is: is there any way to configure it so wan1 being used as the only interface unless its bandwidth reaches some predetermined bw threshold (i.e. 70 Mbps) and keeps being so for the next, say, 20 seconds so the upcoming connections go through wan2? This would be a 1 1/2 bandwidth increase (in my case) with multi-connection services (i.e. torrents, multi-client downloads, simultaneous internet usage of different services by different LAN members etc).