diff --git a/piker/ui/_axes.py b/piker/ui/_axes.py
index 46f64b5d..24389f77 100644
--- a/piker/ui/_axes.py
+++ b/piker/ui/_axes.py
@@ -1,8 +1,6 @@
 """
 Chart axes graphics and behavior.
 """
-import time
-from functools import partial
 from typing import List
 
 
@@ -12,7 +10,6 @@ import pyqtgraph as pg
 from PyQt5 import QtCore, QtGui
 from PyQt5.QtCore import QPointF
 
-from .quantdom.utils import fromtimestamp
 from ._style import _font, hcolor
 
 
@@ -78,7 +75,8 @@ class DynamicDateAxis(pg.AxisItem):
         bars = self.linked_charts.chart._array
         times = bars['time']
         bars_len = len(bars)
-        delay = times[-1] - times[times != times[-1]][-1]
+        # delay = times[-1] - times[times != times[-1]][-1]
+        delay = times[-1] - times[-2]
 
         epochs = times[list(
             map(int, filter(lambda i: i < bars_len, indexes))
@@ -87,7 +85,6 @@ class DynamicDateAxis(pg.AxisItem):
         dts = pd.to_datetime(epochs, unit='s') - 4*pd.offsets.Hour()
         return dts.strftime(self.tick_tpl[delay])
 
-
     def tickStrings(self, values: List[float], scale, spacing):
         return self._indexes_to_timestrs(values)
 
diff --git a/piker/ui/_chart.py b/piker/ui/_chart.py
index dd100fb9..13071f1a 100644
--- a/piker/ui/_chart.py
+++ b/piker/ui/_chart.py
@@ -22,7 +22,7 @@ from .. import brokers
 from .. import data
 from ..log import get_logger
 from ._exec import run_qtractor
-from ._source import ohlc_dtype
+from ._source import base_ohlc_dtype
 from ._interaction import ChartView
 from .. import fsp
 
@@ -507,13 +507,13 @@ class ChartPlotWidget(pg.PlotWidget):
         # TODO: should probably just have some kinda attr mark
         # that determines this behavior based on array type
         try:
-            ylow = bars['low'].min()
-            yhigh = bars['high'].max()
+            ylow = np.nanmin(bars['low'])
+            yhigh = np.nanmax(bars['high'])
             # std = np.std(bars['close'])
         except IndexError:
             # must be non-ohlc array?
-            ylow = bars.min()
-            yhigh = bars.max()
+            ylow = np.nanmin(bars)
+            yhigh = np.nanmax(bars)
             # std = np.std(bars)
 
         # view margins: stay within 10% of the "true range"
@@ -589,14 +589,13 @@ async def add_new_bars(delay_s, linked_charts):
 
         def incr_ohlc_array(array: np.ndarray):
             (index, t, close) = array[-1][['index', 'time', 'close']]
-            new_array = np.append(
-                array,
-                np.array(
-                    [(index + 1, t + delay_s, close, close,
-                      close, close, 0)],
-                    dtype=array.dtype
-                ),
-            )
+
+            # this copies non-std fields (eg. vwap) from the last datum
+            _next = np.array(array[-1], dtype=array.dtype)
+            _next[
+                ['index', 'time', 'volume', 'open', 'high', 'low', 'close']
+            ] = (index + 1, t + delay_s, 0, close, close, close, close)
+            new_array = np.append(array, _next,)
             return new_array
 
         # add new increment/bar
@@ -668,6 +667,9 @@ async def _async_main(
         # figure out the exact symbol
         bars = await client.bars(symbol=sym)
 
+    # allow broker to declare historical data fields
+    ohlc_dtype = getattr(brokermod, 'ohlc_dtype', base_ohlc_dtype)
+
     # remember, msgpack-numpy's ``from_buffer` returns read-only array
     bars = np.array(bars[list(ohlc_dtype.names)])
 
diff --git a/piker/ui/_source.py b/piker/ui/_source.py
index 0224538b..b8816636 100644
--- a/piker/ui/_source.py
+++ b/piker/ui/_source.py
@@ -8,7 +8,7 @@ import numpy as np
 import pandas as pd
 
 
-ohlc_dtype = np.dtype(
+base_ohlc_dtype = np.dtype(
     [
         ('index', int),
         ('time', float),
@@ -38,7 +38,7 @@ def ohlc_zeros(length: int) -> np.ndarray:
     For "why a structarray" see here: https://stackoverflow.com/a/52443038
     Bottom line, they're faster then ``np.recarray``.
     """
-    return np.zeros(length, dtype=ohlc_dtype)
+    return np.zeros(length, dtype=base_ohlc_dtype)
 
 
 @dataclass
@@ -88,7 +88,7 @@ def from_df(
     df = df.rename(columns=columns)
 
     for name in df.columns:
-        if name not in ohlc_dtype.names[1:]:
+        if name not in base_ohlc_dtype.names[1:]:
             del df[name]
 
     # TODO: it turns out column access on recarrays is actually slower: